[
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10413.0,
    "tokens_per_success": 10413.0,
    "conditional_tokens_among_solved_loops": 10413.0,
    "mean_tokens_per_solved_task": 10413.0,
    "mean_input_tokens_per_repair_loop": 9378.0,
    "mean_output_tokens_per_repair_loop": 1035.0,
    "mean_cache_read_tokens_per_repair_loop": 6095.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.272025,
    "mean_cost_per_repair_loop": 0.090675,
    "p95_cost_per_repair_loop": 0.093055,
    "cpsc": 0.090675,
    "realized_cpsc": 0.090675,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.272025,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.090675,
    "mean_spend_per_solved_task": 0.090675,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0988625,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.008187499999999995,
    "cost_delta_vs_gateway_ratio": -0.08281704387406748
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 9149.0,
    "tokens_per_success": 9149.0,
    "conditional_tokens_among_solved_loops": 9149.0,
    "mean_tokens_per_solved_task": 9149.0,
    "mean_input_tokens_per_repair_loop": 8450.333333333334,
    "mean_output_tokens_per_repair_loop": 698.6666666666666,
    "mean_cache_read_tokens_per_repair_loop": 5982.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.333333333333333,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.333333333333333,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.196796,
    "mean_cost_per_repair_loop": 0.06559866666666667,
    "p95_cost_per_repair_loop": 0.067396,
    "cpsc": 0.06559866666666667,
    "realized_cpsc": 0.06559866666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.196796,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.06559866666666667,
    "mean_spend_per_solved_task": 0.06559866666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07174783333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0061491666666666596,
    "cost_delta_vs_gateway_ratio": -0.0857052593922696
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10670.333333333334,
    "tokens_per_success": 10670.333333333334,
    "conditional_tokens_among_solved_loops": 10670.333333333334,
    "mean_tokens_per_solved_task": 10670.333333333334,
    "mean_input_tokens_per_repair_loop": 9565.0,
    "mean_output_tokens_per_repair_loop": 1105.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 6197.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.285431,
    "mean_cost_per_repair_loop": 0.09514366666666667,
    "p95_cost_per_repair_loop": 0.096341,
    "cpsc": 0.09514366666666667,
    "realized_cpsc": 0.09514366666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.285431,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09514366666666667,
    "mean_spend_per_solved_task": 0.09514366666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.10354366666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0084,
    "cost_delta_vs_gateway_ratio": -0.08112519355762947
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 8581.666666666666,
    "tokens_per_success": 8581.666666666666,
    "conditional_tokens_among_solved_loops": 8581.666666666666,
    "mean_tokens_per_solved_task": 8581.666666666666,
    "mean_input_tokens_per_repair_loop": 7781.0,
    "mean_output_tokens_per_repair_loop": 800.6666666666666,
    "mean_cache_read_tokens_per_repair_loop": 5223.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.21249099999999999,
    "mean_cost_per_repair_loop": 0.07083033333333333,
    "p95_cost_per_repair_loop": 0.075331,
    "cpsc": 0.07083033333333333,
    "realized_cpsc": 0.07083033333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.21249099999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07083033333333333,
    "mean_spend_per_solved_task": 0.07083033333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07720366666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006373333333333332,
    "cost_delta_vs_gateway_ratio": -0.08255221038724413
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 16826.666666666668,
    "tokens_per_success": 16826.666666666668,
    "conditional_tokens_among_solved_loops": 16826.666666666668,
    "mean_tokens_per_solved_task": 16826.666666666668,
    "mean_input_tokens_per_repair_loop": 16027.333333333334,
    "mean_output_tokens_per_repair_loop": 799.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 11970.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 5.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.277512,
    "mean_cost_per_repair_loop": 0.09250399999999999,
    "p95_cost_per_repair_loop": 0.116836,
    "cpsc": 0.09250399999999999,
    "realized_cpsc": 0.09250399999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.277512,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09250399999999999,
    "mean_spend_per_solved_task": 0.09250399999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.102619,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.010115000000000004,
    "cost_delta_vs_gateway_ratio": -0.09856849121507717
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 29382.333333333332,
    "tokens_per_success": 29382.333333333332,
    "conditional_tokens_among_solved_loops": 29382.333333333332,
    "mean_tokens_per_solved_task": 29382.333333333332,
    "mean_input_tokens_per_repair_loop": 27300.666666666668,
    "mean_output_tokens_per_repair_loop": 2081.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 19989.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.591558,
    "mean_cost_per_repair_loop": 0.197186,
    "p95_cost_per_repair_loop": 0.204436,
    "cpsc": 0.197186,
    "realized_cpsc": 0.197186,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.591558,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.197186,
    "mean_spend_per_solved_task": 0.197186,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2154326666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.01824666666666667,
    "cost_delta_vs_gateway_ratio": -0.0846977709972799
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 8588.333333333334,
    "tokens_per_success": 8588.333333333334,
    "conditional_tokens_among_solved_loops": 8588.333333333334,
    "mean_tokens_per_solved_task": 8588.333333333334,
    "mean_input_tokens_per_repair_loop": 7955.333333333333,
    "mean_output_tokens_per_repair_loop": 633.0,
    "mean_cache_read_tokens_per_repair_loop": 5348.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 3.6666666666666665,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 3.6666666666666665,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.189196,
    "mean_cost_per_repair_loop": 0.06306533333333333,
    "p95_cost_per_repair_loop": 0.073419,
    "cpsc": 0.06306533333333333,
    "realized_cpsc": 0.06306533333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.189196,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.06306533333333333,
    "mean_spend_per_solved_task": 0.06306533333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06956366666666668,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006498333333333338,
    "cost_delta_vs_gateway_ratio": -0.09341562405661964
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 9445.333333333334,
    "tokens_per_success": 9445.333333333334,
    "conditional_tokens_among_solved_loops": 9445.333333333334,
    "mean_tokens_per_solved_task": 9445.333333333334,
    "mean_input_tokens_per_repair_loop": 8709.0,
    "mean_output_tokens_per_repair_loop": 736.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 6008.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.3333333333333333,
    "mean_verifier_submissions_to_success": 1.3333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.209486,
    "mean_cost_per_repair_loop": 0.06982866666666666,
    "p95_cost_per_repair_loop": 0.106818,
    "cpsc": 0.06982866666666666,
    "realized_cpsc": 0.06982866666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.209486,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.06982866666666666,
    "mean_spend_per_solved_task": 0.06982866666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0765595,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006730833333333334,
    "cost_delta_vs_gateway_ratio": -0.08791637005640494
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 18682.0,
    "tokens_per_success": 18682.0,
    "conditional_tokens_among_solved_loops": 18682.0,
    "mean_tokens_per_solved_task": 18682.0,
    "mean_input_tokens_per_repair_loop": 17465.0,
    "mean_output_tokens_per_repair_loop": 1217.0,
    "mean_cache_read_tokens_per_repair_loop": 9976.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.333333333333333,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.333333333333333,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.43712999999999996,
    "mean_cost_per_repair_loop": 0.14570999999999998,
    "p95_cost_per_repair_loop": 0.189932,
    "cpsc": 0.14570999999999998,
    "realized_cpsc": 0.14570999999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.43712999999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14570999999999998,
    "mean_spend_per_solved_task": 0.14570999999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.16440416666666668,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.018694166666666696,
    "cost_delta_vs_gateway_ratio": -0.11370859416580091
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 40545.666666666664,
    "tokens_per_success": 40545.666666666664,
    "conditional_tokens_among_solved_loops": 40545.666666666664,
    "mean_tokens_per_solved_task": 40545.666666666664,
    "mean_input_tokens_per_repair_loop": 36363.0,
    "mean_output_tokens_per_repair_loop": 4182.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 26012.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.015957,
    "mean_cost_per_repair_loop": 0.33865233333333333,
    "p95_cost_per_repair_loop": 0.372533,
    "cpsc": 0.33865233333333333,
    "realized_cpsc": 0.33865233333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.015957,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.33865233333333333,
    "mean_spend_per_solved_task": 0.33865233333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.36449899999999996,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.025846666666666646,
    "cost_delta_vs_gateway_ratio": -0.07091011680873377
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 13121.333333333334,
    "tokens_per_success": 13121.333333333334,
    "conditional_tokens_among_solved_loops": 13121.333333333334,
    "mean_tokens_per_solved_task": 13121.333333333334,
    "mean_input_tokens_per_repair_loop": 11741.0,
    "mean_output_tokens_per_repair_loop": 1380.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 8121.333333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.666666666666667,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.666666666666667,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.340004,
    "mean_cost_per_repair_loop": 0.11333466666666665,
    "p95_cost_per_repair_loop": 0.136513,
    "cpsc": 0.11333466666666665,
    "realized_cpsc": 0.11333466666666665,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.340004,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11333466666666665,
    "mean_spend_per_solved_task": 0.11333466666666665,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12236049999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.009025833333333325,
    "cost_delta_vs_gateway_ratio": -0.07376427305652826
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 18062.0,
    "tokens_per_success": 18062.0,
    "conditional_tokens_among_solved_loops": 18062.0,
    "mean_tokens_per_solved_task": 18062.0,
    "mean_input_tokens_per_repair_loop": 16687.333333333332,
    "mean_output_tokens_per_repair_loop": 1374.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 11561.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.666666666666667,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.666666666666667,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.394673,
    "mean_cost_per_repair_loop": 0.13155766666666666,
    "p95_cost_per_repair_loop": 0.145669,
    "cpsc": 0.13155766666666666,
    "realized_cpsc": 0.13155766666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.394673,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.13155766666666666,
    "mean_spend_per_solved_task": 0.13155766666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14435016666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.012792500000000004,
    "cost_delta_vs_gateway_ratio": -0.08862130398186818
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 7162.666666666667,
    "tokens_per_success": 7162.666666666667,
    "conditional_tokens_among_solved_loops": 7162.666666666667,
    "mean_tokens_per_solved_task": 7162.666666666667,
    "mean_input_tokens_per_repair_loop": 6550.333333333333,
    "mean_output_tokens_per_repair_loop": 612.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 3901.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 3.0,
    "p95_turns_per_repair_loop": 3.0,
    "mean_agent_steps_per_repair_loop": 3.0,
    "p95_agent_steps_per_repair_loop": 3.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.183033,
    "mean_cost_per_repair_loop": 0.061011,
    "p95_cost_per_repair_loop": 0.06152,
    "cpsc": 0.061011,
    "realized_cpsc": 0.061011,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.183033,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.061011,
    "mean_spend_per_solved_task": 0.061011,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06761933333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006608333333333337,
    "cost_delta_vs_gateway_ratio": -0.09772846030228045
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 8729.0,
    "tokens_per_success": 8729.0,
    "conditional_tokens_among_solved_loops": 8729.0,
    "mean_tokens_per_solved_task": 8729.0,
    "mean_input_tokens_per_repair_loop": 8227.666666666666,
    "mean_output_tokens_per_repair_loop": 501.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 5523.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.172891,
    "mean_cost_per_repair_loop": 0.05763033333333333,
    "p95_cost_per_repair_loop": 0.058648,
    "cpsc": 0.05763033333333333,
    "realized_cpsc": 0.05763033333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.172891,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.05763033333333333,
    "mean_spend_per_solved_task": 0.05763033333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06437033333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0067400000000000055,
    "cost_delta_vs_gateway_ratio": -0.10470661950898714
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 9927.666666666666,
    "tokens_per_success": 9927.666666666666,
    "conditional_tokens_among_solved_loops": 9927.666666666666,
    "mean_tokens_per_solved_task": 9927.666666666666,
    "mean_input_tokens_per_repair_loop": 9244.333333333334,
    "mean_output_tokens_per_repair_loop": 683.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 6462.333333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.333333333333333,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.333333333333333,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.205347,
    "mean_cost_per_repair_loop": 0.068449,
    "p95_cost_per_repair_loop": 0.073625,
    "cpsc": 0.068449,
    "realized_cpsc": 0.068449,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.205347,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.068449,
    "mean_spend_per_solved_task": 0.068449,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07538233333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006933333333333329,
    "cost_delta_vs_gateway_ratio": -0.09197557341021542
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 30986.333333333332,
    "tokens_per_success": 30986.333333333332,
    "conditional_tokens_among_solved_loops": 30986.333333333332,
    "mean_tokens_per_solved_task": 30986.333333333332,
    "mean_input_tokens_per_repair_loop": 27980.333333333332,
    "mean_output_tokens_per_repair_loop": 3006.0,
    "mean_cache_read_tokens_per_repair_loop": 16928.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.833236,
    "mean_cost_per_repair_loop": 0.27774533333333334,
    "p95_cost_per_repair_loop": 0.330053,
    "cpsc": 0.27774533333333334,
    "realized_cpsc": 0.27774533333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.833236,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.27774533333333334,
    "mean_spend_per_solved_task": 0.27774533333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.30534616666666664,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.027600833333333335,
    "cost_delta_vs_gateway_ratio": -0.0903919431333945
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10934.666666666666,
    "tokens_per_success": 10934.666666666666,
    "conditional_tokens_among_solved_loops": 10934.666666666666,
    "mean_tokens_per_solved_task": 10934.666666666666,
    "mean_input_tokens_per_repair_loop": 10263.333333333334,
    "mean_output_tokens_per_repair_loop": 671.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 7395.333333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.333333333333333,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.333333333333333,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.208926,
    "mean_cost_per_repair_loop": 0.069642,
    "p95_cost_per_repair_loop": 0.074741,
    "cpsc": 0.069642,
    "realized_cpsc": 0.069642,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.208926,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.069642,
    "mean_spend_per_solved_task": 0.069642,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07679033333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.00714833333333333,
    "cost_delta_vs_gateway_ratio": -0.09308897387257942
  },
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 14963.0,
    "tokens_per_success": 14963.0,
    "conditional_tokens_among_solved_loops": 14963.0,
    "mean_tokens_per_solved_task": 14963.0,
    "mean_input_tokens_per_repair_loop": 13695.0,
    "mean_output_tokens_per_repair_loop": 1268.0,
    "mean_cache_read_tokens_per_repair_loop": 10245.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.3333333333333333,
    "mean_verifier_submissions_to_success": 1.3333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.324426,
    "mean_cost_per_repair_loop": 0.108142,
    "p95_cost_per_repair_loop": 0.180134,
    "cpsc": 0.108142,
    "realized_cpsc": 0.108142,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.324426,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.108142,
    "mean_spend_per_solved_task": 0.108142,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11673783333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.008595833333333339,
    "cost_delta_vs_gateway_ratio": -0.07363365489908304
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 11732.0,
    "tokens_per_success": 11732.0,
    "conditional_tokens_among_solved_loops": 11732.0,
    "mean_tokens_per_solved_task": 11732.0,
    "mean_input_tokens_per_repair_loop": 10382.666666666666,
    "mean_output_tokens_per_repair_loop": 1349.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 4339.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1983635,
    "mean_cost_per_repair_loop": 0.06612116666666666,
    "p95_cost_per_repair_loop": 0.088125,
    "cpsc": 0.06612116666666666,
    "realized_cpsc": 0.06612116666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1983635,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.06612116666666666,
    "mean_spend_per_solved_task": 0.06612116666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07366575,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.007544583333333336,
    "cost_delta_vs_gateway_ratio": -0.1024164327836659
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 12910.0,
    "tokens_per_success": 12910.0,
    "conditional_tokens_among_solved_loops": 12910.0,
    "mean_tokens_per_solved_task": 12910.0,
    "mean_input_tokens_per_repair_loop": 11810.333333333334,
    "mean_output_tokens_per_repair_loop": 1099.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 5961.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 5.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1791565,
    "mean_cost_per_repair_loop": 0.05971883333333333,
    "p95_cost_per_repair_loop": 0.0967125,
    "cpsc": 0.05971883333333333,
    "realized_cpsc": 0.05971883333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1791565,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.05971883333333333,
    "mean_spend_per_solved_task": 0.05971883333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06701800000000001,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.007299166666666672,
    "cost_delta_vs_gateway_ratio": -0.10891352571945852
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 32302.0,
    "tokens_per_success": 32302.0,
    "conditional_tokens_among_solved_loops": 32302.0,
    "mean_tokens_per_solved_task": 32302.0,
    "mean_input_tokens_per_repair_loop": 30291.666666666668,
    "mean_output_tokens_per_repair_loop": 2010.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 23810.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.283715,
    "mean_cost_per_repair_loop": 0.09457166666666666,
    "p95_cost_per_repair_loop": 0.104105,
    "cpsc": 0.09457166666666666,
    "realized_cpsc": 0.09457166666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.283715,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09457166666666666,
    "mean_spend_per_solved_task": 0.09457166666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.10265625,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.00808458333333334,
    "cost_delta_vs_gateway_ratio": -0.07875393201420604
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 36664.333333333336,
    "tokens_per_success": 36664.333333333336,
    "conditional_tokens_among_solved_loops": 36664.333333333336,
    "mean_tokens_per_solved_task": 36664.333333333336,
    "mean_input_tokens_per_repair_loop": 34668.666666666664,
    "mean_output_tokens_per_repair_loop": 1995.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 29784.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 12.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 12.0,
    "mean_verifier_submissions_per_repair_loop": 2.0,
    "mean_verifier_submissions_to_success": 2.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.267612,
    "mean_cost_per_repair_loop": 0.089204,
    "p95_cost_per_repair_loop": 0.0992785,
    "cpsc": 0.089204,
    "realized_cpsc": 0.089204,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.267612,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.089204,
    "mean_spend_per_solved_task": 0.089204,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.09528233333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006078333333333315,
    "cost_delta_vs_gateway_ratio": -0.06379286821271501
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 22483.0,
    "tokens_per_success": 22483.0,
    "conditional_tokens_among_solved_loops": 22483.0,
    "mean_tokens_per_solved_task": 22483.0,
    "mean_input_tokens_per_repair_loop": 21300.0,
    "mean_output_tokens_per_repair_loop": 1183.0,
    "mean_cache_read_tokens_per_repair_loop": 16930.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.17966100000000002,
    "mean_cost_per_repair_loop": 0.059887,
    "p95_cost_per_repair_loop": 0.0667505,
    "cpsc": 0.059887,
    "realized_cpsc": 0.059887,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.17966100000000002,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.059887,
    "mean_spend_per_solved_task": 0.059887,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06533116666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.005444166666666657,
    "cost_delta_vs_gateway_ratio": -0.08333184518874337
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 46626.333333333336,
    "tokens_per_success": 46626.333333333336,
    "conditional_tokens_among_solved_loops": 46626.333333333336,
    "mean_tokens_per_solved_task": 46626.333333333336,
    "mean_input_tokens_per_repair_loop": 43241.333333333336,
    "mean_output_tokens_per_repair_loop": 3385.0,
    "mean_cache_read_tokens_per_repair_loop": 34563.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.43588550000000004,
    "mean_cost_per_repair_loop": 0.14529516666666667,
    "p95_cost_per_repair_loop": 0.175746,
    "cpsc": 0.14529516666666667,
    "realized_cpsc": 0.14529516666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.43588550000000004,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14529516666666667,
    "mean_spend_per_solved_task": 0.14529516666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.15612225,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.010827083333333301,
    "cost_delta_vs_gateway_ratio": -0.069350033921067
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 9984.0,
    "tokens_per_success": 9984.0,
    "conditional_tokens_among_solved_loops": 9984.0,
    "mean_tokens_per_solved_task": 9984.0,
    "mean_input_tokens_per_repair_loop": 9092.666666666666,
    "mean_output_tokens_per_repair_loop": 891.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 6096.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.120944,
    "mean_cost_per_repair_loop": 0.040314666666666665,
    "p95_cost_per_repair_loop": 0.040713,
    "cpsc": 0.040314666666666665,
    "realized_cpsc": 0.040314666666666665,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.120944,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.040314666666666665,
    "mean_spend_per_solved_task": 0.040314666666666665,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0440505,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0037358333333333362,
    "cost_delta_vs_gateway_ratio": -0.08480796661407558
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 9386.333333333334,
    "tokens_per_success": 9386.333333333334,
    "conditional_tokens_among_solved_loops": 9386.333333333334,
    "mean_tokens_per_solved_task": 9386.333333333334,
    "mean_input_tokens_per_repair_loop": 8691.333333333334,
    "mean_output_tokens_per_repair_loop": 695.0,
    "mean_cache_read_tokens_per_repair_loop": 5923.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.10252549999999999,
    "mean_cost_per_repair_loop": 0.034175166666666666,
    "p95_cost_per_repair_loop": 0.0347365,
    "cpsc": 0.034175166666666666,
    "realized_cpsc": 0.034175166666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.10252549999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.034175166666666666,
    "mean_spend_per_solved_task": 0.034175166666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.037624750000000005,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0034495833333333392,
    "cost_delta_vs_gateway_ratio": -0.09168388715761139
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 40118.333333333336,
    "tokens_per_success": 40118.333333333336,
    "conditional_tokens_among_solved_loops": 40118.333333333336,
    "mean_tokens_per_solved_task": 40118.333333333336,
    "mean_input_tokens_per_repair_loop": 38376.0,
    "mean_output_tokens_per_repair_loop": 1742.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 32242.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.333333333333334,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 9.333333333333334,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.271048,
    "mean_cost_per_repair_loop": 0.09034933333333334,
    "p95_cost_per_repair_loop": 0.0979935,
    "cpsc": 0.09034933333333334,
    "realized_cpsc": 0.09034933333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.271048,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09034933333333334,
    "mean_spend_per_solved_task": 0.09034933333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.09799350000000001,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.007644166666666674,
    "cost_delta_vs_gateway_ratio": -0.07800687460562868
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 68752.66666666667,
    "tokens_per_success": 68752.66666666667,
    "conditional_tokens_among_solved_loops": 68752.66666666667,
    "mean_tokens_per_solved_task": 68752.66666666667,
    "mean_input_tokens_per_repair_loop": 63447.666666666664,
    "mean_output_tokens_per_repair_loop": 5305.0,
    "mean_cache_read_tokens_per_repair_loop": 51621.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.6526975,
    "mean_cost_per_repair_loop": 0.21756583333333335,
    "p95_cost_per_repair_loop": 0.235563,
    "cpsc": 0.21756583333333335,
    "realized_cpsc": 0.21756583333333335,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.6526975,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.21756583333333335,
    "mean_spend_per_solved_task": 0.21756583333333335,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.23232833333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.014762499999999975,
    "cost_delta_vs_gateway_ratio": -0.06354153963141233
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 21094.333333333332,
    "tokens_per_success": 21094.333333333332,
    "conditional_tokens_among_solved_loops": 21094.333333333332,
    "mean_tokens_per_solved_task": 21094.333333333332,
    "mean_input_tokens_per_repair_loop": 19365.666666666668,
    "mean_output_tokens_per_repair_loop": 1728.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 14074.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.230127,
    "mean_cost_per_repair_loop": 0.076709,
    "p95_cost_per_repair_loop": 0.1100585,
    "cpsc": 0.076709,
    "realized_cpsc": 0.076709,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.230127,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.076709,
    "mean_spend_per_solved_task": 0.076709,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08330858333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006599583333333335,
    "cost_delta_vs_gateway_ratio": -0.07921852790278715
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 35703.333333333336,
    "tokens_per_success": 35703.333333333336,
    "conditional_tokens_among_solved_loops": 35703.333333333336,
    "mean_tokens_per_solved_task": 35703.333333333336,
    "mean_input_tokens_per_repair_loop": 32822.333333333336,
    "mean_output_tokens_per_repair_loop": 2881.0,
    "mean_cache_read_tokens_per_repair_loop": 25123.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.6666666666666667,
    "mean_verifier_submissions_to_success": 1.6666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.36924049999999997,
    "mean_cost_per_repair_loop": 0.12308016666666666,
    "p95_cost_per_repair_loop": 0.1439495,
    "cpsc": 0.12308016666666666,
    "realized_cpsc": 0.12308016666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.36924049999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12308016666666666,
    "mean_spend_per_solved_task": 0.12308016666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.13268683333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.009606666666666689,
    "cost_delta_vs_gateway_ratio": -0.07240105461355766
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10782.0,
    "tokens_per_success": 10782.0,
    "conditional_tokens_among_solved_loops": 10782.0,
    "mean_tokens_per_solved_task": 10782.0,
    "mean_input_tokens_per_repair_loop": 10027.666666666666,
    "mean_output_tokens_per_repair_loop": 754.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 6766.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.11563999999999999,
    "mean_cost_per_repair_loop": 0.03854666666666667,
    "p95_cost_per_repair_loop": 0.038943,
    "cpsc": 0.03854666666666667,
    "realized_cpsc": 0.03854666666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.11563999999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.03854666666666667,
    "mean_spend_per_solved_task": 0.03854666666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04261291666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.004066250000000003,
    "cost_delta_vs_gateway_ratio": -0.09542294492084762
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 9584.0,
    "tokens_per_success": 9584.0,
    "conditional_tokens_among_solved_loops": 9584.0,
    "mean_tokens_per_solved_task": 9584.0,
    "mean_input_tokens_per_repair_loop": 8980.333333333334,
    "mean_output_tokens_per_repair_loop": 603.6666666666666,
    "mean_cache_read_tokens_per_repair_loop": 5993.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0990655,
    "mean_cost_per_repair_loop": 0.033021833333333334,
    "p95_cost_per_repair_loop": 0.0343735,
    "cpsc": 0.033021833333333334,
    "realized_cpsc": 0.033021833333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0990655,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.033021833333333334,
    "mean_spend_per_solved_task": 0.033021833333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03674516666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0037233333333333333,
    "cost_delta_vs_gateway_ratio": -0.10132851939710891
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10502.333333333334,
    "tokens_per_success": 10502.333333333334,
    "conditional_tokens_among_solved_loops": 10502.333333333334,
    "mean_tokens_per_solved_task": 10502.333333333334,
    "mean_input_tokens_per_repair_loop": 9764.0,
    "mean_output_tokens_per_repair_loop": 738.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 6214.333333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.666666666666667,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.666666666666667,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1179415,
    "mean_cost_per_repair_loop": 0.03931383333333333,
    "p95_cost_per_repair_loop": 0.047512,
    "cpsc": 0.03931383333333333,
    "realized_cpsc": 0.03931383333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1179415,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.03931383333333333,
    "mean_spend_per_solved_task": 0.03931383333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04373925,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0044254166666666634,
    "cost_delta_vs_gateway_ratio": -0.10117724164604247
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 46394.666666666664,
    "tokens_per_success": 46394.666666666664,
    "conditional_tokens_among_solved_loops": 46394.666666666664,
    "mean_tokens_per_solved_task": 46394.666666666664,
    "mean_input_tokens_per_repair_loop": 43244.666666666664,
    "mean_output_tokens_per_repair_loop": 3150.0,
    "mean_cache_read_tokens_per_repair_loop": 32227.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.449851,
    "mean_cost_per_repair_loop": 0.14995033333333332,
    "p95_cost_per_repair_loop": 0.1654195,
    "cpsc": 0.14995033333333332,
    "realized_cpsc": 0.14995033333333332,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.449851,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14995033333333332,
    "mean_spend_per_solved_task": 0.14995033333333332,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1637036666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.013753333333333359,
    "cost_delta_vs_gateway_ratio": -0.084013593668234
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10723.333333333334,
    "tokens_per_success": 10723.333333333334,
    "conditional_tokens_among_solved_loops": 10723.333333333334,
    "mean_tokens_per_solved_task": 10723.333333333334,
    "mean_input_tokens_per_repair_loop": 9956.666666666666,
    "mean_output_tokens_per_repair_loop": 766.6666666666666,
    "mean_cache_read_tokens_per_repair_loop": 6772.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.115428,
    "mean_cost_per_repair_loop": 0.038476,
    "p95_cost_per_repair_loop": 0.03938,
    "cpsc": 0.038476,
    "realized_cpsc": 0.038476,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.115428,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.038476,
    "mean_spend_per_solved_task": 0.038476,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04244683333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0039708333333333306,
    "cost_delta_vs_gateway_ratio": -0.09354839976284052
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10953.0,
    "tokens_per_success": 10953.0,
    "conditional_tokens_among_solved_loops": 10953.0,
    "mean_tokens_per_solved_task": 10953.0,
    "mean_input_tokens_per_repair_loop": 10007.0,
    "mean_output_tokens_per_repair_loop": 946.0,
    "mean_cache_read_tokens_per_repair_loop": 6518.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.133053,
    "mean_cost_per_repair_loop": 0.044351,
    "p95_cost_per_repair_loop": 0.044886,
    "cpsc": 0.044351,
    "realized_cpsc": 0.044351,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.133053,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.044351,
    "mean_spend_per_solved_task": 0.044351,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.048701416666666664,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.004350416666666662,
    "cost_delta_vs_gateway_ratio": -0.08932833918246723
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17092.0,
    "tokens_per_success": 17092.0,
    "conditional_tokens_among_solved_loops": 17092.0,
    "mean_tokens_per_solved_task": 17092.0,
    "mean_input_tokens_per_repair_loop": 15216.333333333334,
    "mean_output_tokens_per_repair_loop": 1875.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 9320.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.24309550000000002,
    "mean_cost_per_repair_loop": 0.08103183333333334,
    "p95_cost_per_repair_loop": 0.0851875,
    "cpsc": 0.08103183333333334,
    "realized_cpsc": 0.08103183333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.24309550000000002,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.08103183333333334,
    "mean_spend_per_solved_task": 0.08103183333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08838933333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.007357499999999999,
    "cost_delta_vs_gateway_ratio": -0.08323968201291254
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 27102.666666666668,
    "tokens_per_success": 27102.666666666668,
    "conditional_tokens_among_solved_loops": 27102.666666666668,
    "mean_tokens_per_solved_task": 27102.666666666668,
    "mean_input_tokens_per_repair_loop": 24596.333333333332,
    "mean_output_tokens_per_repair_loop": 2506.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 17116.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.3258495,
    "mean_cost_per_repair_loop": 0.1086165,
    "p95_cost_per_repair_loop": 0.140233,
    "cpsc": 0.1086165,
    "realized_cpsc": 0.1086165,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.3258495,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1086165,
    "mean_spend_per_solved_task": 0.1086165,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1179465,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.009329999999999986,
    "cost_delta_vs_gateway_ratio": -0.07910366140580675
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 35343.666666666664,
    "tokens_per_success": 35343.666666666664,
    "conditional_tokens_among_solved_loops": 35343.666666666664,
    "mean_tokens_per_solved_task": 35343.666666666664,
    "mean_input_tokens_per_repair_loop": 32428.0,
    "mean_output_tokens_per_repair_loop": 2915.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 25722.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.3578435,
    "mean_cost_per_repair_loop": 0.11928116666666666,
    "p95_cost_per_repair_loop": 0.147023,
    "cpsc": 0.11928116666666666,
    "realized_cpsc": 0.11928116666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.3578435,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11928116666666666,
    "mean_spend_per_solved_task": 0.11928116666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12764491666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.008363750000000003,
    "cost_delta_vs_gateway_ratio": -0.06552356504600329
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 34741.333333333336,
    "tokens_per_success": 34741.333333333336,
    "conditional_tokens_among_solved_loops": 34741.333333333336,
    "mean_tokens_per_solved_task": 34741.333333333336,
    "mean_input_tokens_per_repair_loop": 31779.333333333332,
    "mean_output_tokens_per_repair_loop": 2962.0,
    "mean_cache_read_tokens_per_repair_loop": 26164.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.0,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 9.0,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.34562149999999997,
    "mean_cost_per_repair_loop": 0.11520716666666665,
    "p95_cost_per_repair_loop": 0.150773,
    "cpsc": 0.11520716666666665,
    "realized_cpsc": 0.11520716666666665,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.34562149999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11520716666666665,
    "mean_spend_per_solved_task": 0.11520716666666665,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12220341666666668,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006996250000000019,
    "cost_delta_vs_gateway_ratio": -0.05725085427916992
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 47441.0,
    "tokens_per_success": 47441.0,
    "conditional_tokens_among_solved_loops": 47441.0,
    "mean_tokens_per_solved_task": 47441.0,
    "mean_input_tokens_per_repair_loop": 45028.333333333336,
    "mean_output_tokens_per_repair_loop": 2412.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 37276.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 9.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.3531445,
    "mean_cost_per_repair_loop": 0.11771483333333332,
    "p95_cost_per_repair_loop": 0.1312235,
    "cpsc": 0.11771483333333332,
    "realized_cpsc": 0.11771483333333332,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.3531445,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11771483333333332,
    "mean_spend_per_solved_task": 0.11771483333333332,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12738233333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.009667499999999996,
    "cost_delta_vs_gateway_ratio": -0.07589356975195406
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 105853.66666666667,
    "tokens_per_success": 105853.66666666667,
    "conditional_tokens_among_solved_loops": 105853.66666666667,
    "mean_tokens_per_solved_task": 105853.66666666667,
    "mean_input_tokens_per_repair_loop": 100417.66666666667,
    "mean_output_tokens_per_repair_loop": 5436.0,
    "mean_cache_read_tokens_per_repair_loop": 88015.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 13.333333333333334,
    "p95_turns_per_repair_loop": 16.0,
    "mean_agent_steps_per_repair_loop": 13.333333333333334,
    "p95_agent_steps_per_repair_loop": 16.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.7257625,
    "mean_cost_per_repair_loop": 0.24192083333333333,
    "p95_cost_per_repair_loop": 0.293546,
    "cpsc": 0.24192083333333333,
    "realized_cpsc": 0.24192083333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.7257625,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.24192083333333333,
    "mean_spend_per_solved_task": 0.24192083333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2573908333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.015469999999999984,
    "cost_delta_vs_gateway_ratio": -0.0601031505265986
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 21763.666666666668,
    "tokens_per_success": 21763.666666666668,
    "conditional_tokens_among_solved_loops": 21763.666666666668,
    "mean_tokens_per_solved_task": 21763.666666666668,
    "mean_input_tokens_per_repair_loop": 19789.666666666668,
    "mean_output_tokens_per_repair_loop": 1974.0,
    "mean_cache_read_tokens_per_repair_loop": 15003.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.24235,
    "mean_cost_per_repair_loop": 0.08078333333333333,
    "p95_cost_per_repair_loop": 0.105403,
    "cpsc": 0.08078333333333333,
    "realized_cpsc": 0.08078333333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.24235,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.08078333333333333,
    "mean_spend_per_solved_task": 0.08078333333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08675041666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.005967083333333327,
    "cost_delta_vs_gateway_ratio": -0.06878449190926068
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 12691.666666666666,
    "tokens_per_success": 12691.666666666666,
    "conditional_tokens_among_solved_loops": 12691.666666666666,
    "mean_tokens_per_solved_task": 12691.666666666666,
    "mean_input_tokens_per_repair_loop": 11643.333333333334,
    "mean_output_tokens_per_repair_loop": 1048.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 8246.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.666666666666667,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.666666666666667,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.141954,
    "mean_cost_per_repair_loop": 0.047318,
    "p95_cost_per_repair_loop": 0.0522405,
    "cpsc": 0.047318,
    "realized_cpsc": 0.047318,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.141954,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.047318,
    "mean_spend_per_solved_task": 0.047318,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.051552999999999995,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.004234999999999998,
    "cost_delta_vs_gateway_ratio": -0.08214846856632976
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 51272.666666666664,
    "tokens_per_success": 51272.666666666664,
    "conditional_tokens_among_solved_loops": 51272.666666666664,
    "mean_tokens_per_solved_task": 51272.666666666664,
    "mean_input_tokens_per_repair_loop": 48874.666666666664,
    "mean_output_tokens_per_repair_loop": 2398.0,
    "mean_cache_read_tokens_per_repair_loop": 34619.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.44560449999999996,
    "mean_cost_per_repair_loop": 0.1485348333333333,
    "p95_cost_per_repair_loop": 0.203294,
    "cpsc": 0.1485348333333333,
    "realized_cpsc": 0.1485348333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.44560449999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1485348333333333,
    "mean_spend_per_solved_task": 0.1485348333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.16632691666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.017792083333333337,
    "cost_delta_vs_gateway_ratio": -0.10697055948551125
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 97696.66666666667,
    "tokens_per_success": 97696.66666666667,
    "conditional_tokens_among_solved_loops": 97696.66666666667,
    "mean_tokens_per_solved_task": 97696.66666666667,
    "mean_input_tokens_per_repair_loop": 89351.66666666667,
    "mean_output_tokens_per_repair_loop": 8345.0,
    "mean_cache_read_tokens_per_repair_loop": 74094.33333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.333333333333334,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 9.333333333333334,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.9658765,
    "mean_cost_per_repair_loop": 0.32195883333333336,
    "p95_cost_per_repair_loop": 0.330992,
    "cpsc": 0.32195883333333336,
    "realized_cpsc": 0.32195883333333336,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.9658765,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.32195883333333336,
    "mean_spend_per_solved_task": 0.32195883333333336,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.3410071666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.019048333333333334,
    "cost_delta_vs_gateway_ratio": -0.05585904108564678
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 57146.0,
    "tokens_per_success": 57146.0,
    "conditional_tokens_among_solved_loops": 57146.0,
    "mean_tokens_per_solved_task": 57146.0,
    "mean_input_tokens_per_repair_loop": 51999.0,
    "mean_output_tokens_per_repair_loop": 5147.0,
    "mean_cache_read_tokens_per_repair_loop": 42466.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.0,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 9.0,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.3333333333333333,
    "mean_verifier_submissions_to_success": 1.3333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5927145,
    "mean_cost_per_repair_loop": 0.1975715,
    "p95_cost_per_repair_loop": 0.3041255,
    "cpsc": 0.1975715,
    "realized_cpsc": 0.1975715,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5927145,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1975715,
    "mean_spend_per_solved_task": 0.1975715,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.20946483333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.011893333333333311,
    "cost_delta_vs_gateway_ratio": -0.056779618535808214
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 29202.666666666668,
    "tokens_per_success": 29202.666666666668,
    "conditional_tokens_among_solved_loops": 29202.666666666668,
    "mean_tokens_per_solved_task": 29202.666666666668,
    "mean_input_tokens_per_repair_loop": 26260.333333333332,
    "mean_output_tokens_per_repair_loop": 2942.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 18558.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.333333333333333,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.333333333333333,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.3640425,
    "mean_cost_per_repair_loop": 0.1213475,
    "p95_cost_per_repair_loop": 0.1353045,
    "cpsc": 0.1213475,
    "realized_cpsc": 0.1213475,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.3640425,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1213475,
    "mean_spend_per_solved_task": 0.1213475,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.13096166666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.009614166666666665,
    "cost_delta_vs_gateway_ratio": -0.07341206714432975
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15350.333333333334,
    "tokens_per_success": 15350.333333333334,
    "conditional_tokens_among_solved_loops": 15350.333333333334,
    "mean_tokens_per_solved_task": 15350.333333333334,
    "mean_input_tokens_per_repair_loop": 14094.666666666666,
    "mean_output_tokens_per_repair_loop": 1255.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 10252.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.0,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 5.0,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.16719299999999998,
    "mean_cost_per_repair_loop": 0.055730999999999996,
    "p95_cost_per_repair_loop": 0.065592,
    "cpsc": 0.055730999999999996,
    "realized_cpsc": 0.055730999999999996,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.16719299999999998,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.055730999999999996,
    "mean_spend_per_solved_task": 0.055730999999999996,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06052183333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.004790833333333337,
    "cost_delta_vs_gateway_ratio": -0.07915876088794407
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 21377.666666666668,
    "tokens_per_success": 21377.666666666668,
    "conditional_tokens_among_solved_loops": 21377.666666666668,
    "mean_tokens_per_solved_task": 21377.666666666668,
    "mean_input_tokens_per_repair_loop": 20104.666666666668,
    "mean_output_tokens_per_repair_loop": 1273.0,
    "mean_cache_read_tokens_per_repair_loop": 15864.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.182872,
    "mean_cost_per_repair_loop": 0.060957333333333336,
    "p95_cost_per_repair_loop": 0.07336,
    "cpsc": 0.060957333333333336,
    "realized_cpsc": 0.060957333333333336,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.182872,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.060957333333333336,
    "mean_spend_per_solved_task": 0.060957333333333336,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06624066666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.005283333333333334,
    "cost_delta_vs_gateway_ratio": -0.07975966425458682
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 21244.333333333332,
    "tokens_per_success": 21244.333333333332,
    "conditional_tokens_among_solved_loops": 21244.333333333332,
    "mean_tokens_per_solved_task": 21244.333333333332,
    "mean_input_tokens_per_repair_loop": 19833.333333333332,
    "mean_output_tokens_per_repair_loop": 1411.0,
    "mean_cache_read_tokens_per_repair_loop": 13899.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2156795,
    "mean_cost_per_repair_loop": 0.07189316666666666,
    "p95_cost_per_repair_loop": 0.0838425,
    "cpsc": 0.07189316666666666,
    "realized_cpsc": 0.07189316666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2156795,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07189316666666666,
    "mean_spend_per_solved_task": 0.07189316666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07929191666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.00739875,
    "cost_delta_vs_gateway_ratio": -0.0933102680706209
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 67800.33333333333,
    "tokens_per_success": 67800.33333333333,
    "conditional_tokens_among_solved_loops": 67800.33333333333,
    "mean_tokens_per_solved_task": 67800.33333333333,
    "mean_input_tokens_per_repair_loop": 62385.666666666664,
    "mean_output_tokens_per_repair_loop": 5414.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 49858.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.6687974999999999,
    "mean_cost_per_repair_loop": 0.22293249999999998,
    "p95_cost_per_repair_loop": 0.2357305,
    "cpsc": 0.22293249999999998,
    "realized_cpsc": 0.22293249999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.6687974999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.22293249999999998,
    "mean_spend_per_solved_task": 0.22293249999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.23857000000000003,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.015637500000000044,
    "cost_delta_vs_gateway_ratio": -0.0655467996814354
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 12702.666666666666,
    "tokens_per_success": 12702.666666666666,
    "conditional_tokens_among_solved_loops": 12702.666666666666,
    "mean_tokens_per_solved_task": 12702.666666666666,
    "mean_input_tokens_per_repair_loop": 11703.0,
    "mean_output_tokens_per_repair_loop": 999.6666666666666,
    "mean_cache_read_tokens_per_repair_loop": 8155.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.333333333333333,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.333333333333333,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1404275,
    "mean_cost_per_repair_loop": 0.04680916666666667,
    "p95_cost_per_repair_loop": 0.0547395,
    "cpsc": 0.04680916666666667,
    "realized_cpsc": 0.04680916666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1404275,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.04680916666666667,
    "mean_spend_per_solved_task": 0.04680916666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.05123333333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.004424166666666664,
    "cost_delta_vs_gateway_ratio": -0.08635328562134022
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 19748.333333333332,
    "tokens_per_success": 19748.333333333332,
    "conditional_tokens_among_solved_loops": 19748.333333333332,
    "mean_tokens_per_solved_task": 19748.333333333332,
    "mean_input_tokens_per_repair_loop": 17876.666666666668,
    "mean_output_tokens_per_repair_loop": 1871.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 13578.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.225222,
    "mean_cost_per_repair_loop": 0.075074,
    "p95_cost_per_repair_loop": 0.0771455,
    "cpsc": 0.075074,
    "realized_cpsc": 0.075074,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.225222,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.075074,
    "mean_spend_per_solved_task": 0.075074,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08043233333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.005358333333333336,
    "cost_delta_vs_gateway_ratio": -0.06661914570011233
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 13462.666666666666,
    "tokens_per_success": 13462.666666666666,
    "conditional_tokens_among_solved_loops": 13462.666666666666,
    "mean_tokens_per_solved_task": 13462.666666666666,
    "mean_input_tokens_per_repair_loop": 12018.333333333334,
    "mean_output_tokens_per_repair_loop": 1444.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 7652.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.666666666666667,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.666666666666667,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1111734,
    "mean_cost_per_repair_loop": 0.0370578,
    "p95_cost_per_repair_loop": 0.0401997,
    "cpsc": 0.0370578,
    "realized_cpsc": 0.0370578,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1111734,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0370578,
    "mean_spend_per_solved_task": 0.0370578,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.02688336666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.010174433333333335,
    "cost_delta_vs_gateway_ratio": 0.37846574275791356
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23478.666666666668,
    "tokens_per_success": 23478.666666666668,
    "conditional_tokens_among_solved_loops": 23478.666666666668,
    "mean_tokens_per_solved_task": 23478.666666666668,
    "mean_input_tokens_per_repair_loop": 22447.0,
    "mean_output_tokens_per_repair_loop": 1031.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 17324.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.10812089999999999,
    "mean_cost_per_repair_loop": 0.0360403,
    "p95_cost_per_repair_loop": 0.0509079,
    "cpsc": 0.0360403,
    "realized_cpsc": 0.0360403,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.10812089999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0360403,
    "mean_spend_per_solved_task": 0.0360403,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.026581866666666665,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.00945843333333333,
    "cost_delta_vs_gateway_ratio": 0.35582276639714283
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23268.666666666668,
    "tokens_per_success": 23268.666666666668,
    "conditional_tokens_among_solved_loops": 23268.666666666668,
    "mean_tokens_per_solved_task": 23268.666666666668,
    "mean_input_tokens_per_repair_loop": 21428.333333333332,
    "mean_output_tokens_per_repair_loop": 1840.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 16888.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.3333333333333333,
    "mean_verifier_submissions_to_success": 1.3333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1388745,
    "mean_cost_per_repair_loop": 0.046291500000000006,
    "p95_cost_per_repair_loop": 0.0581949,
    "cpsc": 0.046291500000000006,
    "realized_cpsc": 0.046291500000000006,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1388745,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.046291500000000006,
    "mean_spend_per_solved_task": 0.046291500000000006,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03312433333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.013167166666666674,
    "cost_delta_vs_gateway_ratio": 0.3975073712175342
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 176034.0,
    "tokens_per_success": 176034.0,
    "conditional_tokens_among_solved_loops": 176034.0,
    "mean_tokens_per_solved_task": 176034.0,
    "mean_input_tokens_per_repair_loop": 172070.66666666666,
    "mean_output_tokens_per_repair_loop": 3963.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 158494.33333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 23.333333333333332,
    "p95_turns_per_repair_loop": 44.0,
    "mean_agent_steps_per_repair_loop": 23.333333333333332,
    "p95_agent_steps_per_repair_loop": 44.0,
    "mean_verifier_submissions_per_repair_loop": 5.333333333333333,
    "mean_verifier_submissions_to_success": 5.333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.4431819,
    "mean_cost_per_repair_loop": 0.1477273,
    "p95_cost_per_repair_loop": 0.28988040000000004,
    "cpsc": 0.1477273,
    "realized_cpsc": 0.1477273,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.4431819,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1477273,
    "mean_spend_per_solved_task": 0.1477273,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1052497,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.042477600000000004,
    "cost_delta_vs_gateway_ratio": 0.4035887988279302
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 19294.666666666668,
    "tokens_per_success": 19294.666666666668,
    "conditional_tokens_among_solved_loops": 19294.666666666668,
    "mean_tokens_per_solved_task": 19294.666666666668,
    "mean_input_tokens_per_repair_loop": 18487.333333333332,
    "mean_output_tokens_per_repair_loop": 807.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 14247.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.08731259999999999,
    "mean_cost_per_repair_loop": 0.029104199999999997,
    "p95_cost_per_repair_loop": 0.0341967,
    "cpsc": 0.029104199999999997,
    "realized_cpsc": 0.029104199999999997,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.08731259999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.029104199999999997,
    "mean_spend_per_solved_task": 0.029104199999999997,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0215168,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0075873999999999985,
    "cost_delta_vs_gateway_ratio": 0.3526267846519928
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 67742.0,
    "tokens_per_success": 67742.0,
    "conditional_tokens_among_solved_loops": 67742.0,
    "mean_tokens_per_solved_task": 67742.0,
    "mean_input_tokens_per_repair_loop": 65046.333333333336,
    "mean_output_tokens_per_repair_loop": 2695.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 55188.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.0,
    "p95_turns_per_repair_loop": 12.0,
    "mean_agent_steps_per_repair_loop": 10.0,
    "p95_agent_steps_per_repair_loop": 12.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2596992,
    "mean_cost_per_repair_loop": 0.0865664,
    "p95_cost_per_repair_loop": 0.0899247,
    "cpsc": 0.0865664,
    "realized_cpsc": 0.0865664,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2596992,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0865664,
    "mean_spend_per_solved_task": 0.0865664,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06263010000000001,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.023936300000000004,
    "cost_delta_vs_gateway_ratio": 0.3821852431977596
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10043.0,
    "tokens_per_success": 10043.0,
    "conditional_tokens_among_solved_loops": 10043.0,
    "mean_tokens_per_solved_task": 10043.0,
    "mean_input_tokens_per_repair_loop": 9162.666666666666,
    "mean_output_tokens_per_repair_loop": 880.3333333333334,
    "mean_cache_read_tokens_per_repair_loop": 6222.333333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0716781,
    "mean_cost_per_repair_loop": 0.0238927,
    "p95_cost_per_repair_loop": 0.026856900000000003,
    "cpsc": 0.0238927,
    "realized_cpsc": 0.0238927,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0716781,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0238927,
    "mean_spend_per_solved_task": 0.0238927,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.017394633333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.006498066666666666,
    "cost_delta_vs_gateway_ratio": 0.37356732632095335
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 9452.666666666666,
    "tokens_per_success": 9452.666666666666,
    "conditional_tokens_among_solved_loops": 9452.666666666666,
    "mean_tokens_per_solved_task": 9452.666666666666,
    "mean_input_tokens_per_repair_loop": 8639.666666666666,
    "mean_output_tokens_per_repair_loop": 813.0,
    "mean_cache_read_tokens_per_repair_loop": 5836.333333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06706770000000001,
    "mean_cost_per_repair_loop": 0.0223559,
    "p95_cost_per_repair_loop": 0.0226041,
    "cpsc": 0.0223559,
    "realized_cpsc": 0.0223559,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06706770000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0223559,
    "mean_spend_per_solved_task": 0.0223559,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0163016,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.006054300000000003,
    "cost_delta_vs_gateway_ratio": 0.3713929920989353
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 33094.333333333336,
    "tokens_per_success": 33094.333333333336,
    "conditional_tokens_among_solved_loops": 33094.333333333336,
    "mean_tokens_per_solved_task": 33094.333333333336,
    "mean_input_tokens_per_repair_loop": 31548.0,
    "mean_output_tokens_per_repair_loop": 1546.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 25793.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.144591,
    "mean_cost_per_repair_loop": 0.048197,
    "p95_cost_per_repair_loop": 0.054056099999999996,
    "cpsc": 0.048197,
    "realized_cpsc": 0.048197,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.144591,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.048197,
    "mean_spend_per_solved_task": 0.048197,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03500066666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.013196333333333329,
    "cost_delta_vs_gateway_ratio": 0.3770309136968818
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 67143.33333333333,
    "tokens_per_success": 67143.33333333333,
    "conditional_tokens_among_solved_loops": 67143.33333333333,
    "mean_tokens_per_solved_task": 67143.33333333333,
    "mean_input_tokens_per_repair_loop": 61648.333333333336,
    "mean_output_tokens_per_repair_loop": 5495.0,
    "mean_cache_read_tokens_per_repair_loop": 45771.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.4313595,
    "mean_cost_per_repair_loop": 0.1437865,
    "p95_cost_per_repair_loop": 0.1609584,
    "cpsc": 0.1437865,
    "realized_cpsc": 0.1437865,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.4313595,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1437865,
    "mean_spend_per_solved_task": 0.1437865,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.10378766666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.03999883333333334,
    "cost_delta_vs_gateway_ratio": 0.38539100663855375
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 18014.0,
    "tokens_per_success": 18014.0,
    "conditional_tokens_among_solved_loops": 18014.0,
    "mean_tokens_per_solved_task": 18014.0,
    "mean_input_tokens_per_repair_loop": 16166.0,
    "mean_output_tokens_per_repair_loop": 1848.0,
    "mean_cache_read_tokens_per_repair_loop": 11700.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1338813,
    "mean_cost_per_repair_loop": 0.0446271,
    "p95_cost_per_repair_loop": 0.0513915,
    "cpsc": 0.0446271,
    "realized_cpsc": 0.0446271,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1338813,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0446271,
    "mean_spend_per_solved_task": 0.0446271,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.031978566666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.012648533333333337,
    "cost_delta_vs_gateway_ratio": 0.3955315904298401
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 52997.333333333336,
    "tokens_per_success": 52997.333333333336,
    "conditional_tokens_among_solved_loops": 52997.333333333336,
    "mean_tokens_per_solved_task": 52997.333333333336,
    "mean_input_tokens_per_repair_loop": 50188.0,
    "mean_output_tokens_per_repair_loop": 2809.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 42858.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 17.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 17.0,
    "mean_verifier_submissions_per_repair_loop": 2.0,
    "mean_verifier_submissions_to_success": 2.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.23095949999999998,
    "mean_cost_per_repair_loop": 0.0769865,
    "p95_cost_per_repair_loop": 0.1363113,
    "cpsc": 0.0769865,
    "realized_cpsc": 0.0769865,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.23095949999999998,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0769865,
    "mean_spend_per_solved_task": 0.0769865,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.05497950000000001,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.02200699999999999,
    "cost_delta_vs_gateway_ratio": 0.4002764666830362
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 14265.0,
    "tokens_per_success": 14265.0,
    "conditional_tokens_among_solved_loops": 14265.0,
    "mean_tokens_per_solved_task": 14265.0,
    "mean_input_tokens_per_repair_loop": 13317.333333333334,
    "mean_output_tokens_per_repair_loop": 947.6666666666666,
    "mean_cache_read_tokens_per_repair_loop": 10031.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.333333333333333,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.333333333333333,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0812445,
    "mean_cost_per_repair_loop": 0.027081499999999998,
    "p95_cost_per_repair_loop": 0.029502599999999997,
    "cpsc": 0.027081499999999998,
    "realized_cpsc": 0.027081499999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0812445,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.027081499999999998,
    "mean_spend_per_solved_task": 0.027081499999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.019691833333333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.007389666666666665,
    "cost_delta_vs_gateway_ratio": 0.3752655500165042
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 18712.0,
    "tokens_per_success": 18712.0,
    "conditional_tokens_among_solved_loops": 18712.0,
    "mean_tokens_per_solved_task": 18712.0,
    "mean_input_tokens_per_repair_loop": 17829.0,
    "mean_output_tokens_per_repair_loop": 883.0,
    "mean_cache_read_tokens_per_repair_loop": 12689.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.09741240000000001,
    "mean_cost_per_repair_loop": 0.0324708,
    "p95_cost_per_repair_loop": 0.0354534,
    "cpsc": 0.0324708,
    "realized_cpsc": 0.0324708,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.09741240000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0324708,
    "mean_spend_per_solved_task": 0.0324708,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.024210033333333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.008260766666666669,
    "cost_delta_vs_gateway_ratio": 0.3412125275884242
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 13453.0,
    "tokens_per_success": 13453.0,
    "conditional_tokens_among_solved_loops": 13453.0,
    "mean_tokens_per_solved_task": 13453.0,
    "mean_input_tokens_per_repair_loop": 12607.0,
    "mean_output_tokens_per_repair_loop": 846.0,
    "mean_cache_read_tokens_per_repair_loop": 9626.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0735624,
    "mean_cost_per_repair_loop": 0.0245208,
    "p95_cost_per_repair_loop": 0.0250305,
    "cpsc": 0.0245208,
    "realized_cpsc": 0.0245208,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0735624,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0245208,
    "mean_spend_per_solved_task": 0.0245208,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.017832033333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.006688766666666665,
    "cost_delta_vs_gateway_ratio": 0.37509837165699916
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 45259.333333333336,
    "tokens_per_success": 45259.333333333336,
    "conditional_tokens_among_solved_loops": 45259.333333333336,
    "mean_tokens_per_solved_task": 45259.333333333336,
    "mean_input_tokens_per_repair_loop": 42076.0,
    "mean_output_tokens_per_repair_loop": 3183.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 29010.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.28694759999999997,
    "mean_cost_per_repair_loop": 0.09564919999999999,
    "p95_cost_per_repair_loop": 0.10980569999999999,
    "cpsc": 0.09564919999999999,
    "realized_cpsc": 0.09564919999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.28694759999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09564919999999999,
    "mean_spend_per_solved_task": 0.09564919999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07029113333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.02535806666666666,
    "cost_delta_vs_gateway_ratio": 0.36075768683959186
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10443.0,
    "tokens_per_success": 10443.0,
    "conditional_tokens_among_solved_loops": 10443.0,
    "mean_tokens_per_solved_task": 10443.0,
    "mean_input_tokens_per_repair_loop": 9571.0,
    "mean_output_tokens_per_repair_loop": 872.0,
    "mean_cache_read_tokens_per_repair_loop": 6508.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0726588,
    "mean_cost_per_repair_loop": 0.024219599999999997,
    "p95_cost_per_repair_loop": 0.0285228,
    "cpsc": 0.024219599999999997,
    "realized_cpsc": 0.024219599999999997,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0726588,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.024219599999999997,
    "mean_spend_per_solved_task": 0.024219599999999997,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.017673566666666668,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.00654603333333333,
    "cost_delta_vs_gateway_ratio": 0.37038552867087743
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 601089.6666666666,
    "tokens_per_success": 601089.6666666666,
    "conditional_tokens_among_solved_loops": 601089.6666666666,
    "mean_tokens_per_solved_task": 601089.6666666666,
    "mean_input_tokens_per_repair_loop": 590945.6666666666,
    "mean_output_tokens_per_repair_loop": 10144.0,
    "mean_cache_read_tokens_per_repair_loop": 572306.3333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 27.0,
    "p95_turns_per_repair_loop": 67.0,
    "mean_agent_steps_per_repair_loop": 27.0,
    "p95_agent_steps_per_repair_loop": 67.0,
    "mean_verifier_submissions_per_repair_loop": 7.0,
    "mean_verifier_submissions_to_success": 7.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.1393096999999999,
    "mean_cost_per_repair_loop": 0.37976989999999994,
    "p95_cost_per_repair_loop": 1.0678857,
    "cpsc": 0.37976989999999994,
    "realized_cpsc": 0.37976989999999994,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.1393096999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.37976989999999994,
    "mean_spend_per_solved_task": 0.37976989999999994,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2624726,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.11729729999999994,
    "cost_delta_vs_gateway_ratio": 0.44689350431244984
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 30013.0,
    "tokens_per_success": 30013.0,
    "conditional_tokens_among_solved_loops": 30013.0,
    "mean_tokens_per_solved_task": 30013.0,
    "mean_input_tokens_per_repair_loop": 27832.666666666668,
    "mean_output_tokens_per_repair_loop": 2180.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 16375.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2159661,
    "mean_cost_per_repair_loop": 0.0719887,
    "p95_cost_per_repair_loop": 0.0872082,
    "cpsc": 0.0719887,
    "realized_cpsc": 0.0719887,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2159661,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0719887,
    "mean_spend_per_solved_task": 0.0719887,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.05371363333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.018275066666666662,
    "cost_delta_vs_gateway_ratio": 0.3402314372080582
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 38759.666666666664,
    "tokens_per_success": 38759.666666666664,
    "conditional_tokens_among_solved_loops": 38759.666666666664,
    "mean_tokens_per_solved_task": 38759.666666666664,
    "mean_input_tokens_per_repair_loop": 37235.666666666664,
    "mean_output_tokens_per_repair_loop": 1524.0,
    "mean_cache_read_tokens_per_repair_loop": 25344.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.19841189999999997,
    "mean_cost_per_repair_loop": 0.0661373,
    "p95_cost_per_repair_loop": 0.124332,
    "cpsc": 0.0661373,
    "realized_cpsc": 0.0661373,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.19841189999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0661373,
    "mean_spend_per_solved_task": 0.0661373,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.050028866666666665,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.016108433333333328,
    "cost_delta_vs_gateway_ratio": 0.3219827752777396
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 58676.333333333336,
    "tokens_per_success": 58676.333333333336,
    "conditional_tokens_among_solved_loops": 58676.333333333336,
    "mean_tokens_per_solved_task": 58676.333333333336,
    "mean_input_tokens_per_repair_loop": 55382.666666666664,
    "mean_output_tokens_per_repair_loop": 3293.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 43822.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 11.666666666666666,
    "p95_turns_per_repair_loop": 16.0,
    "mean_agent_steps_per_repair_loop": 11.666666666666666,
    "p95_agent_steps_per_repair_loop": 16.0,
    "mean_verifier_submissions_per_repair_loop": 1.6666666666666667,
    "mean_verifier_submissions_to_success": 1.6666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2916981,
    "mean_cost_per_repair_loop": 0.0972327,
    "p95_cost_per_repair_loop": 0.1207902,
    "cpsc": 0.0972327,
    "realized_cpsc": 0.0972327,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2916981,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0972327,
    "mean_spend_per_solved_task": 0.0972327,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0705903,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.02664240000000001,
    "cost_delta_vs_gateway_ratio": 0.3774229603784091
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 87434.0,
    "tokens_per_success": 87434.0,
    "conditional_tokens_among_solved_loops": 87434.0,
    "mean_tokens_per_solved_task": 87434.0,
    "mean_input_tokens_per_repair_loop": 82955.33333333333,
    "mean_output_tokens_per_repair_loop": 4478.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 67871.33333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 15.333333333333334,
    "p95_turns_per_repair_loop": 22.0,
    "mean_agent_steps_per_repair_loop": 15.333333333333334,
    "p95_agent_steps_per_repair_loop": 22.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.3983802,
    "mean_cost_per_repair_loop": 0.1327934,
    "p95_cost_per_repair_loop": 0.1970574,
    "cpsc": 0.1327934,
    "realized_cpsc": 0.1327934,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.3983802,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1327934,
    "mean_spend_per_solved_task": 0.1327934,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0960556,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.03673780000000001,
    "cost_delta_vs_gateway_ratio": 0.38246390632092253
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 42083.333333333336,
    "tokens_per_success": 42083.333333333336,
    "conditional_tokens_among_solved_loops": 42083.333333333336,
    "mean_tokens_per_solved_task": 42083.333333333336,
    "mean_input_tokens_per_repair_loop": 40431.333333333336,
    "mean_output_tokens_per_repair_loop": 1652.0,
    "mean_cache_read_tokens_per_repair_loop": 34574.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 15.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 15.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1581726,
    "mean_cost_per_repair_loop": 0.0527242,
    "p95_cost_per_repair_loop": 0.0869751,
    "cpsc": 0.0527242,
    "realized_cpsc": 0.0527242,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1581726,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0527242,
    "mean_spend_per_solved_task": 0.0527242,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03806846666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.014655733333333329,
    "cost_delta_vs_gateway_ratio": 0.38498354718778605
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 82239.33333333333,
    "tokens_per_success": 82239.33333333333,
    "conditional_tokens_among_solved_loops": 82239.33333333333,
    "mean_tokens_per_solved_task": 82239.33333333333,
    "mean_input_tokens_per_repair_loop": 79034.33333333333,
    "mean_output_tokens_per_repair_loop": 3205.0,
    "mean_cache_read_tokens_per_repair_loop": 63430.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 11.0,
    "p95_turns_per_repair_loop": 13.0,
    "mean_agent_steps_per_repair_loop": 11.0,
    "p95_agent_steps_per_repair_loop": 13.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.3417483,
    "mean_cost_per_repair_loop": 0.1139161,
    "p95_cost_per_repair_loop": 0.1184802,
    "cpsc": 0.1139161,
    "realized_cpsc": 0.1139161,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.3417483,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1139161,
    "mean_spend_per_solved_task": 0.1139161,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08373506666666665,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.030181033333333347,
    "cost_delta_vs_gateway_ratio": 0.3604348158397997
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 16748.0,
    "tokens_per_success": 16748.0,
    "conditional_tokens_among_solved_loops": 16748.0,
    "mean_tokens_per_solved_task": 16748.0,
    "mean_input_tokens_per_repair_loop": 15554.0,
    "mean_output_tokens_per_repair_loop": 1194.0,
    "mean_cache_read_tokens_per_repair_loop": 11968.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.09677250000000001,
    "mean_cost_per_repair_loop": 0.0322575,
    "p95_cost_per_repair_loop": 0.0340449,
    "cpsc": 0.0322575,
    "realized_cpsc": 0.0322575,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.09677250000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0322575,
    "mean_spend_per_solved_task": 0.0322575,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.02329183333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.008965666666666672,
    "cost_delta_vs_gateway_ratio": 0.38492747815758055
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17035.666666666668,
    "tokens_per_success": 17035.666666666668,
    "conditional_tokens_among_solved_loops": 17035.666666666668,
    "mean_tokens_per_solved_task": 17035.666666666668,
    "mean_input_tokens_per_repair_loop": 15841.0,
    "mean_output_tokens_per_repair_loop": 1194.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 12518.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0949305,
    "mean_cost_per_repair_loop": 0.0316435,
    "p95_cost_per_repair_loop": 0.0358491,
    "cpsc": 0.0316435,
    "realized_cpsc": 0.0316435,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0949305,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0316435,
    "mean_spend_per_solved_task": 0.0316435,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.022750666666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.008892833333333336,
    "cost_delta_vs_gateway_ratio": 0.3908823184668583
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 58264.0,
    "tokens_per_success": 58264.0,
    "conditional_tokens_among_solved_loops": 58264.0,
    "mean_tokens_per_solved_task": 58264.0,
    "mean_input_tokens_per_repair_loop": 55926.666666666664,
    "mean_output_tokens_per_repair_loop": 2337.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 42828.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 11.333333333333334,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 11.333333333333334,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.26161049999999997,
    "mean_cost_per_repair_loop": 0.08720349999999999,
    "p95_cost_per_repair_loop": 0.0956973,
    "cpsc": 0.08720349999999999,
    "realized_cpsc": 0.08720349999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.26161049999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.08720349999999999,
    "mean_spend_per_solved_task": 0.08720349999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0646735,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.022529999999999994,
    "cost_delta_vs_gateway_ratio": 0.3483652500637819
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 116897.66666666667,
    "tokens_per_success": 116897.66666666667,
    "conditional_tokens_among_solved_loops": 116897.66666666667,
    "mean_tokens_per_solved_task": 116897.66666666667,
    "mean_input_tokens_per_repair_loop": 108557.66666666667,
    "mean_output_tokens_per_repair_loop": 8340.0,
    "mean_cache_read_tokens_per_repair_loop": 90043.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 12.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 12.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.6229707,
    "mean_cost_per_repair_loop": 0.2076569,
    "p95_cost_per_repair_loop": 0.2148,
    "cpsc": 0.2076569,
    "realized_cpsc": 0.2076569,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.6229707,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.2076569,
    "mean_spend_per_solved_task": 0.2076569,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14768459999999997,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.05997230000000001,
    "cost_delta_vs_gateway_ratio": 0.40608364040665057
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 47310.666666666664,
    "tokens_per_success": 47310.666666666664,
    "conditional_tokens_among_solved_loops": 47310.666666666664,
    "mean_tokens_per_solved_task": 47310.666666666664,
    "mean_input_tokens_per_repair_loop": 42570.666666666664,
    "mean_output_tokens_per_repair_loop": 4740.0,
    "mean_cache_read_tokens_per_repair_loop": 34463.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.3172857,
    "mean_cost_per_repair_loop": 0.1057619,
    "p95_cost_per_repair_loop": 0.18492270000000002,
    "cpsc": 0.1057619,
    "realized_cpsc": 0.1057619,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.3172857,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1057619,
    "mean_spend_per_solved_task": 0.1057619,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0745531,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.03120880000000001,
    "cost_delta_vs_gateway_ratio": 0.4186117009218934
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 32574.666666666668,
    "tokens_per_success": 32574.666666666668,
    "conditional_tokens_among_solved_loops": 32574.666666666668,
    "mean_tokens_per_solved_task": 32574.666666666668,
    "mean_input_tokens_per_repair_loop": 30672.0,
    "mean_output_tokens_per_repair_loop": 1902.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 22384.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1803549,
    "mean_cost_per_repair_loop": 0.060118300000000006,
    "p95_cost_per_repair_loop": 0.063228,
    "cpsc": 0.060118300000000006,
    "realized_cpsc": 0.060118300000000006,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1803549,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.060118300000000006,
    "mean_spend_per_solved_task": 0.060118300000000006,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04421503333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.015903266666666676,
    "cost_delta_vs_gateway_ratio": 0.359680078645951
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17615.333333333332,
    "tokens_per_success": 17615.333333333332,
    "conditional_tokens_among_solved_loops": 17615.333333333332,
    "mean_tokens_per_solved_task": 17615.333333333332,
    "mean_input_tokens_per_repair_loop": 16391.0,
    "mean_output_tokens_per_repair_loop": 1224.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 12491.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.10143150000000001,
    "mean_cost_per_repair_loop": 0.0338105,
    "p95_cost_per_repair_loop": 0.035152800000000005,
    "cpsc": 0.0338105,
    "realized_cpsc": 0.0338105,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.10143150000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0338105,
    "mean_spend_per_solved_task": 0.0338105,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.024484000000000002,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.009326500000000001,
    "cost_delta_vs_gateway_ratio": 0.38092223492893323
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 25106.333333333332,
    "tokens_per_success": 25106.333333333332,
    "conditional_tokens_among_solved_loops": 25106.333333333332,
    "mean_tokens_per_solved_task": 25106.333333333332,
    "mean_input_tokens_per_repair_loop": 24175.333333333332,
    "mean_output_tokens_per_repair_loop": 931.0,
    "mean_cache_read_tokens_per_repair_loop": 16293.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1274997,
    "mean_cost_per_repair_loop": 0.0424999,
    "p95_cost_per_repair_loop": 0.060934199999999994,
    "cpsc": 0.0424999,
    "realized_cpsc": 0.0424999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1274997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0424999,
    "mean_spend_per_solved_task": 0.0424999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0322681,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0102318,
    "cost_delta_vs_gateway_ratio": 0.3170871541863326
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 26947.333333333332,
    "tokens_per_success": 26947.333333333332,
    "conditional_tokens_among_solved_loops": 26947.333333333332,
    "mean_tokens_per_solved_task": 26947.333333333332,
    "mean_input_tokens_per_repair_loop": 25535.666666666668,
    "mean_output_tokens_per_repair_loop": 1411.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 17748.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1495818,
    "mean_cost_per_repair_loop": 0.0498606,
    "p95_cost_per_repair_loop": 0.0734937,
    "cpsc": 0.0498606,
    "realized_cpsc": 0.0498606,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1495818,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0498606,
    "mean_spend_per_solved_task": 0.0498606,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03712423333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.012736366666666665,
    "cost_delta_vs_gateway_ratio": 0.34307420040997477
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 76197.0,
    "tokens_per_success": 76197.0,
    "conditional_tokens_among_solved_loops": 76197.0,
    "mean_tokens_per_solved_task": 76197.0,
    "mean_input_tokens_per_repair_loop": 71200.0,
    "mean_output_tokens_per_repair_loop": 4997.0,
    "mean_cache_read_tokens_per_repair_loop": 52065.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.333333333333334,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 10.333333333333334,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.44393309999999997,
    "mean_cost_per_repair_loop": 0.1479777,
    "p95_cost_per_repair_loop": 0.1598814,
    "cpsc": 0.1479777,
    "realized_cpsc": 0.1479777,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.44393309999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1479777,
    "mean_spend_per_solved_task": 0.1479777,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.10820863333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.039769066666666665,
    "cost_delta_vs_gateway_ratio": 0.3675221231577641
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 20827.666666666668,
    "tokens_per_success": 20827.666666666668,
    "conditional_tokens_among_solved_loops": 20827.666666666668,
    "mean_tokens_per_solved_task": 20827.666666666668,
    "mean_input_tokens_per_repair_loop": 19478.333333333332,
    "mean_output_tokens_per_repair_loop": 1349.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 15575.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1098648,
    "mean_cost_per_repair_loop": 0.0366216,
    "p95_cost_per_repair_loop": 0.0426171,
    "cpsc": 0.0366216,
    "realized_cpsc": 0.0366216,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1098648,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0366216,
    "mean_spend_per_solved_task": 0.0366216,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.026359233333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.010262366666666667,
    "cost_delta_vs_gateway_ratio": 0.38932720602647775
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 66247.33333333333,
    "tokens_per_success": 66247.33333333333,
    "conditional_tokens_among_solved_loops": 66247.33333333333,
    "mean_tokens_per_solved_task": 66247.33333333333,
    "mean_input_tokens_per_repair_loop": 63015.0,
    "mean_output_tokens_per_repair_loop": 3232.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 51963.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 12.333333333333334,
    "p95_turns_per_repair_loop": 13.0,
    "mean_agent_steps_per_repair_loop": 12.333333333333334,
    "p95_agent_steps_per_repair_loop": 13.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2916897,
    "mean_cost_per_repair_loop": 0.0972299,
    "p95_cost_per_repair_loop": 0.11984460000000001,
    "cpsc": 0.0972299,
    "realized_cpsc": 0.0972299,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2916897,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0972299,
    "mean_spend_per_solved_task": 0.0972299,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0703336,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0268963,
    "cost_delta_vs_gateway_ratio": 0.382410398443987
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 343839.3333333333,
    "tokens_per_success": 343839.3333333333,
    "conditional_tokens_among_solved_loops": 343839.3333333333,
    "mean_tokens_per_solved_task": 343839.3333333333,
    "mean_input_tokens_per_repair_loop": 323978.0,
    "mean_output_tokens_per_repair_loop": 19861.333333333332,
    "mean_cache_read_tokens_per_repair_loop": 222980.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 26.333333333333332,
    "p95_turns_per_repair_loop": 32.0,
    "mean_agent_steps_per_repair_loop": 26.333333333333332,
    "p95_agent_steps_per_repair_loop": 32.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.091088,
    "mean_cost_per_repair_loop": 0.363696,
    "p95_cost_per_repair_loop": 0.40423215,
    "cpsc": 0.363696,
    "realized_cpsc": 0.363696,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.091088,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.363696,
    "mean_spend_per_solved_task": 0.363696,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.36369599999999996,
    "mean_cost_delta_vs_gateway_per_repair_loop": 7.401486830834377e-17,
    "cost_delta_vs_gateway_ratio": 2.035075126158764e-16
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 99992.33333333333,
    "tokens_per_success": 99992.33333333333,
    "conditional_tokens_among_solved_loops": 99992.33333333333,
    "mean_tokens_per_solved_task": 99992.33333333333,
    "mean_input_tokens_per_repair_loop": 90991.0,
    "mean_output_tokens_per_repair_loop": 9001.333333333334,
    "mean_cache_read_tokens_per_repair_loop": 46073.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 19.0,
    "p95_turns_per_repair_loop": 21.0,
    "mean_agent_steps_per_repair_loop": 19.0,
    "p95_agent_steps_per_repair_loop": 21.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.4658985,
    "mean_cost_per_repair_loop": 0.1552995,
    "p95_cost_per_repair_loop": 0.16121115,
    "cpsc": 0.1552995,
    "realized_cpsc": 0.1552995,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.4658985,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1552995,
    "mean_spend_per_solved_task": 0.1552995,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1552995,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 1886491.6666666667,
    "tokens_per_success": 1886491.6666666667,
    "conditional_tokens_among_solved_loops": 1886491.6666666667,
    "mean_tokens_per_solved_task": 1886491.6666666667,
    "mean_input_tokens_per_repair_loop": 1820612.3333333333,
    "mean_output_tokens_per_repair_loop": 65879.33333333333,
    "mean_cache_read_tokens_per_repair_loop": 1371946.3333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 55.333333333333336,
    "p95_turns_per_repair_loop": 61.0,
    "mean_agent_steps_per_repair_loop": 55.333333333333336,
    "p95_agent_steps_per_repair_loop": 61.0,
    "mean_verifier_submissions_per_repair_loop": 4.0,
    "mean_verifier_submissions_to_success": 4.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 4.41511485,
    "mean_cost_per_repair_loop": 1.4717049500000001,
    "p95_cost_per_repair_loop": 1.58396565,
    "cpsc": 1.4717049500000001,
    "realized_cpsc": 1.4717049500000001,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 4.41511485,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 1.4717049500000001,
    "mean_spend_per_solved_task": 1.4717049500000001,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 1.4386633,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.03304165000000007,
    "cost_delta_vs_gateway_ratio": 0.02296690963062731
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 250901.66666666666,
    "tokens_per_success": 250901.66666666666,
    "conditional_tokens_among_solved_loops": 250901.66666666666,
    "mean_tokens_per_solved_task": 250901.66666666666,
    "mean_input_tokens_per_repair_loop": 233900.33333333334,
    "mean_output_tokens_per_repair_loop": 17001.333333333332,
    "mean_cache_read_tokens_per_repair_loop": 142399.66666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 28.666666666666668,
    "p95_turns_per_repair_loop": 32.0,
    "mean_agent_steps_per_repair_loop": 28.666666666666668,
    "p95_agent_steps_per_repair_loop": 32.0,
    "mean_verifier_submissions_per_repair_loop": 1.6666666666666667,
    "mean_verifier_submissions_to_success": 1.6666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.93486885,
    "mean_cost_per_repair_loop": 0.31162295,
    "p95_cost_per_repair_loop": 0.45766455,
    "cpsc": 0.31162295,
    "realized_cpsc": 0.31162295,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.93486885,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.31162295,
    "mean_spend_per_solved_task": 0.31162295,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.31162295,
    "mean_cost_delta_vs_gateway_per_repair_loop": -3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": -1.187570881867715e-16
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 140706.66666666666,
    "tokens_per_success": 140706.66666666666,
    "conditional_tokens_among_solved_loops": 140706.66666666666,
    "mean_tokens_per_solved_task": 140706.66666666666,
    "mean_input_tokens_per_repair_loop": 134227.0,
    "mean_output_tokens_per_repair_loop": 6479.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 69905.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 25.0,
    "p95_turns_per_repair_loop": 27.0,
    "mean_agent_steps_per_repair_loop": 25.0,
    "p95_agent_steps_per_repair_loop": 27.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.49585725,
    "mean_cost_per_repair_loop": 0.16528575,
    "p95_cost_per_repair_loop": 0.1739379,
    "cpsc": 0.16528575,
    "realized_cpsc": 0.16528575,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.49585725,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.16528575,
    "mean_spend_per_solved_task": 0.16528575,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.16528575,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 358919.3333333333,
    "tokens_per_success": 358919.3333333333,
    "conditional_tokens_among_solved_loops": 358919.3333333333,
    "mean_tokens_per_solved_task": 358919.3333333333,
    "mean_input_tokens_per_repair_loop": 344752.0,
    "mean_output_tokens_per_repair_loop": 14167.333333333334,
    "mean_cache_read_tokens_per_repair_loop": 245847.66666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 34.666666666666664,
    "p95_turns_per_repair_loop": 38.0,
    "mean_agent_steps_per_repair_loop": 34.666666666666664,
    "p95_agent_steps_per_repair_loop": 38.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.93821895,
    "mean_cost_per_repair_loop": 0.31273965,
    "p95_cost_per_repair_loop": 0.36590205,
    "cpsc": 0.31273965,
    "realized_cpsc": 0.31273965,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.93821895,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.31273965,
    "mean_spend_per_solved_task": 0.31273965,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.31273965,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 280815.6666666667,
    "tokens_per_success": 280815.6666666667,
    "conditional_tokens_among_solved_loops": 280815.6666666667,
    "mean_tokens_per_solved_task": 280815.6666666667,
    "mean_input_tokens_per_repair_loop": 264428.0,
    "mean_output_tokens_per_repair_loop": 16387.666666666668,
    "mean_cache_read_tokens_per_repair_loop": 194253.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 26.0,
    "p95_turns_per_repair_loop": 30.0,
    "mean_agent_steps_per_repair_loop": 26.0,
    "p95_agent_steps_per_repair_loop": 30.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.84566835,
    "mean_cost_per_repair_loop": 0.28188945,
    "p95_cost_per_repair_loop": 0.3548577,
    "cpsc": 0.28188945,
    "realized_cpsc": 0.28188945,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.84566835,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.28188945,
    "mean_spend_per_solved_task": 0.28188945,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.28188945000000004,
    "mean_cost_delta_vs_gateway_per_repair_loop": -3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": -1.3128350193372574e-16
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 84356.0,
    "tokens_per_success": 84356.0,
    "conditional_tokens_among_solved_loops": 84356.0,
    "mean_tokens_per_solved_task": 84356.0,
    "mean_input_tokens_per_repair_loop": 77657.0,
    "mean_output_tokens_per_repair_loop": 6699.0,
    "mean_cache_read_tokens_per_repair_loop": 36108.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 18.0,
    "p95_turns_per_repair_loop": 21.0,
    "mean_agent_steps_per_repair_loop": 18.0,
    "p95_agent_steps_per_repair_loop": 21.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.38409075,
    "mean_cost_per_repair_loop": 0.12803024999999998,
    "p95_cost_per_repair_loop": 0.15484755,
    "cpsc": 0.12803024999999998,
    "realized_cpsc": 0.12803024999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.38409075,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12803024999999998,
    "mean_spend_per_solved_task": 0.12803024999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12803025,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.445261340744546e-16
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 89121.66666666667,
    "tokens_per_success": 89121.66666666667,
    "conditional_tokens_among_solved_loops": 89121.66666666667,
    "mean_tokens_per_solved_task": 89121.66666666667,
    "mean_input_tokens_per_repair_loop": 83167.33333333333,
    "mean_output_tokens_per_repair_loop": 5954.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 42377.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 19.666666666666668,
    "p95_turns_per_repair_loop": 22.0,
    "mean_agent_steps_per_repair_loop": 19.666666666666668,
    "p95_agent_steps_per_repair_loop": 22.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.36339045000000003,
    "mean_cost_per_repair_loop": 0.12113015,
    "p95_cost_per_repair_loop": 0.13584525,
    "cpsc": 0.12113015,
    "realized_cpsc": 0.12113015,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.36339045000000003,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12113015,
    "mean_spend_per_solved_task": 0.12113015,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12113014999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": 1.5275897104961849e-16
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 496509.3333333333,
    "tokens_per_success": 496509.3333333333,
    "conditional_tokens_among_solved_loops": 496509.3333333333,
    "mean_tokens_per_solved_task": 496509.3333333333,
    "mean_input_tokens_per_repair_loop": 453810.3333333333,
    "mean_output_tokens_per_repair_loop": 42699.0,
    "mean_cache_read_tokens_per_repair_loop": 318787.3333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 22.333333333333332,
    "p95_turns_per_repair_loop": 26.0,
    "mean_agent_steps_per_repair_loop": 22.333333333333332,
    "p95_agent_steps_per_repair_loop": 26.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.9039308,
    "mean_cost_per_repair_loop": 0.6346436,
    "p95_cost_per_repair_loop": 0.8438611500000001,
    "cpsc": 0.6346436,
    "realized_cpsc": 0.6346436,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.9039308,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.6346436,
    "mean_spend_per_solved_task": 0.6346436,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.47719065,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.15745295,
    "cost_delta_vs_gateway_ratio": 0.32995816242417997
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 278997.0,
    "tokens_per_success": 278997.0,
    "conditional_tokens_among_solved_loops": 278997.0,
    "mean_tokens_per_solved_task": 278997.0,
    "mean_input_tokens_per_repair_loop": 257093.0,
    "mean_output_tokens_per_repair_loop": 21904.0,
    "mean_cache_read_tokens_per_repair_loop": 178999.33333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 18.666666666666668,
    "p95_turns_per_repair_loop": 22.0,
    "mean_agent_steps_per_repair_loop": 18.666666666666668,
    "p95_agent_steps_per_repair_loop": 22.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.0233792,
    "mean_cost_per_repair_loop": 0.3411264,
    "p95_cost_per_repair_loop": 0.46700759999999997,
    "cpsc": 0.3411264,
    "realized_cpsc": 0.3411264,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.0233792,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.3411264,
    "mean_spend_per_solved_task": 0.3411264,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.3411264,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 238842.33333333334,
    "tokens_per_success": 238842.33333333334,
    "conditional_tokens_among_solved_loops": 238842.33333333334,
    "mean_tokens_per_solved_task": 238842.33333333334,
    "mean_input_tokens_per_repair_loop": 214014.0,
    "mean_output_tokens_per_repair_loop": 24828.333333333332,
    "mean_cache_read_tokens_per_repair_loop": 132043.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 19.666666666666668,
    "p95_turns_per_repair_loop": 24.0,
    "mean_agent_steps_per_repair_loop": 19.666666666666668,
    "p95_agent_steps_per_repair_loop": 24.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.09865385,
    "mean_cost_per_repair_loop": 0.36621795,
    "p95_cost_per_repair_loop": 0.4169982,
    "cpsc": 0.36621795,
    "realized_cpsc": 0.36621795,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.09865385,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.36621795,
    "mean_spend_per_solved_task": 0.36621795,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.31333649999999996,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.05288145000000003,
    "cost_delta_vs_gateway_ratio": 0.16876887946345234
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 80387.66666666667,
    "tokens_per_success": 80387.66666666667,
    "conditional_tokens_among_solved_loops": 80387.66666666667,
    "mean_tokens_per_solved_task": 80387.66666666667,
    "mean_input_tokens_per_repair_loop": 74671.33333333333,
    "mean_output_tokens_per_repair_loop": 5716.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 36729.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 17.333333333333332,
    "p95_turns_per_repair_loop": 20.0,
    "mean_agent_steps_per_repair_loop": 17.333333333333332,
    "p95_agent_steps_per_repair_loop": 20.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.34160955,
    "mean_cost_per_repair_loop": 0.11386985,
    "p95_cost_per_repair_loop": 0.13756154999999998,
    "cpsc": 0.11386985,
    "realized_cpsc": 0.11386985,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.34160955,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11386985,
    "mean_spend_per_solved_task": 0.11386985,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11386985000000001,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.624988271881094e-16
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 133758.0,
    "tokens_per_success": 133758.0,
    "conditional_tokens_among_solved_loops": 133758.0,
    "mean_tokens_per_solved_task": 133758.0,
    "mean_input_tokens_per_repair_loop": 124291.0,
    "mean_output_tokens_per_repair_loop": 9467.0,
    "mean_cache_read_tokens_per_repair_loop": 70480.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 21.333333333333332,
    "p95_turns_per_repair_loop": 23.0,
    "mean_agent_steps_per_repair_loop": 21.333333333333332,
    "p95_agent_steps_per_repair_loop": 23.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5294745000000001,
    "mean_cost_per_repair_loop": 0.17649150000000002,
    "p95_cost_per_repair_loop": 0.1820652,
    "cpsc": 0.17649150000000002,
    "realized_cpsc": 0.17649150000000002,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5294745000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.17649150000000002,
    "mean_spend_per_solved_task": 0.17649150000000002,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1764915,
    "mean_cost_delta_vs_gateway_per_repair_loop": 3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": 2.0968394599270724e-16
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 134630.66666666666,
    "tokens_per_success": 134630.66666666666,
    "conditional_tokens_among_solved_loops": 134630.66666666666,
    "mean_tokens_per_solved_task": 134630.66666666666,
    "mean_input_tokens_per_repair_loop": 127168.0,
    "mean_output_tokens_per_repair_loop": 7462.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 65642.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 26.0,
    "p95_turns_per_repair_loop": 30.0,
    "mean_agent_steps_per_repair_loop": 26.0,
    "p95_agent_steps_per_repair_loop": 30.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5078979,
    "mean_cost_per_repair_loop": 0.1692993,
    "p95_cost_per_repair_loop": 0.19405455,
    "cpsc": 0.1692993,
    "realized_cpsc": 0.1692993,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5078979,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1692993,
    "mean_spend_per_solved_task": 0.1692993,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1692993,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 410633.6666666667,
    "tokens_per_success": 410633.6666666667,
    "conditional_tokens_among_solved_loops": 410633.6666666667,
    "mean_tokens_per_solved_task": 410633.6666666667,
    "mean_input_tokens_per_repair_loop": 382465.0,
    "mean_output_tokens_per_repair_loop": 28168.666666666668,
    "mean_cache_read_tokens_per_repair_loop": 271691.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 28.333333333333332,
    "p95_turns_per_repair_loop": 33.0,
    "mean_agent_steps_per_repair_loop": 28.333333333333332,
    "p95_agent_steps_per_repair_loop": 33.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.38129795,
    "mean_cost_per_repair_loop": 0.46043265,
    "p95_cost_per_repair_loop": 0.4726701,
    "cpsc": 0.46043265,
    "realized_cpsc": 0.46043265,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.38129795,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.46043265,
    "mean_spend_per_solved_task": 0.46043265,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.40678220000000004,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.05365044999999998,
    "cost_delta_vs_gateway_ratio": 0.13188986637075067
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 90814.66666666667,
    "tokens_per_success": 90814.66666666667,
    "conditional_tokens_among_solved_loops": 90814.66666666667,
    "mean_tokens_per_solved_task": 90814.66666666667,
    "mean_input_tokens_per_repair_loop": 84648.33333333333,
    "mean_output_tokens_per_repair_loop": 6166.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 45910.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 18.0,
    "p95_turns_per_repair_loop": 22.0,
    "mean_agent_steps_per_repair_loop": 18.0,
    "p95_agent_steps_per_repair_loop": 22.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.36147300000000004,
    "mean_cost_per_repair_loop": 0.12049100000000001,
    "p95_cost_per_repair_loop": 0.14924685000000001,
    "cpsc": 0.12049100000000001,
    "realized_cpsc": 0.12049100000000001,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.36147300000000004,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12049100000000001,
    "mean_spend_per_solved_task": 0.12049100000000001,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12049099999999997,
    "mean_cost_delta_vs_gateway_per_repair_loop": 3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": 3.0713857594485806e-16
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 139829.66666666666,
    "tokens_per_success": 139829.66666666666,
    "conditional_tokens_among_solved_loops": 139829.66666666666,
    "mean_tokens_per_solved_task": 139829.66666666666,
    "mean_input_tokens_per_repair_loop": 131525.0,
    "mean_output_tokens_per_repair_loop": 8304.666666666666,
    "mean_cache_read_tokens_per_repair_loop": 75581.33333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 24.0,
    "p95_turns_per_repair_loop": 25.0,
    "mean_agent_steps_per_repair_loop": 24.0,
    "p95_agent_steps_per_repair_loop": 25.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5099840999999999,
    "mean_cost_per_repair_loop": 0.16999469999999997,
    "p95_cost_per_repair_loop": 0.1891569,
    "cpsc": 0.16999469999999997,
    "realized_cpsc": 0.16999469999999997,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5099840999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.16999469999999997,
    "mean_spend_per_solved_task": 0.16999469999999997,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.16999470000000003,
    "mean_cost_delta_vs_gateway_per_repair_loop": -3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": -2.1769757618426857e-16
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 31791.666666666668,
    "tokens_per_success": 31791.666666666668,
    "conditional_tokens_among_solved_loops": 31791.666666666668,
    "mean_tokens_per_solved_task": 31791.666666666668,
    "mean_input_tokens_per_repair_loop": 29421.333333333332,
    "mean_output_tokens_per_repair_loop": 2370.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 17445.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 12.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 12.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.059325619999999996,
    "mean_cost_per_repair_loop": 0.019775206666666666,
    "p95_cost_per_repair_loop": 0.02981296,
    "cpsc": 0.019775206666666666,
    "realized_cpsc": 0.019775206666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.059325619999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.019775206666666666,
    "mean_spend_per_solved_task": 0.019775206666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.02176846,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0019932533333333335,
    "cost_delta_vs_gateway_ratio": -0.0915661159922812
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 38649.666666666664,
    "tokens_per_success": 38649.666666666664,
    "conditional_tokens_among_solved_loops": 38649.666666666664,
    "mean_tokens_per_solved_task": 38649.666666666664,
    "mean_input_tokens_per_repair_loop": 35735.666666666664,
    "mean_output_tokens_per_repair_loop": 2914.0,
    "mean_cache_read_tokens_per_repair_loop": 24074.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 11.333333333333334,
    "p95_turns_per_repair_loop": 13.0,
    "mean_agent_steps_per_repair_loop": 11.333333333333334,
    "p95_agent_steps_per_repair_loop": 13.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06731802,
    "mean_cost_per_repair_loop": 0.022439340000000002,
    "p95_cost_per_repair_loop": 0.026783599999999998,
    "cpsc": 0.022439340000000002,
    "realized_cpsc": 0.022439340000000002,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06731802,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.022439340000000002,
    "mean_spend_per_solved_task": 0.022439340000000002,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.022883016666666662,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.000443676666666661,
    "cost_delta_vs_gateway_ratio": -0.019388906328637955
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 72622.66666666667,
    "tokens_per_success": 72622.66666666667,
    "conditional_tokens_among_solved_loops": 72622.66666666667,
    "mean_tokens_per_solved_task": 72622.66666666667,
    "mean_input_tokens_per_repair_loop": 66868.66666666667,
    "mean_output_tokens_per_repair_loop": 5754.0,
    "mean_cache_read_tokens_per_repair_loop": 45728.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 15.0,
    "p95_turns_per_repair_loop": 20.0,
    "mean_agent_steps_per_repair_loop": 15.0,
    "p95_agent_steps_per_repair_loop": 20.0,
    "mean_verifier_submissions_per_repair_loop": 2.3333333333333335,
    "mean_verifier_submissions_to_success": 2.3333333333333335,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.12792688,
    "mean_cost_per_repair_loop": 0.04264229333333333,
    "p95_cost_per_repair_loop": 0.060085839999999995,
    "cpsc": 0.04264229333333333,
    "realized_cpsc": 0.04264229333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.12792688,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.04264229333333333,
    "mean_spend_per_solved_task": 0.04264229333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04783523333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.005192940000000007,
    "cost_delta_vs_gateway_ratio": -0.10855889347949259
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 39519.0,
    "tokens_per_success": 39519.0,
    "conditional_tokens_among_solved_loops": 39519.0,
    "mean_tokens_per_solved_task": 39519.0,
    "mean_input_tokens_per_repair_loop": 36830.0,
    "mean_output_tokens_per_repair_loop": 2689.0,
    "mean_cache_read_tokens_per_repair_loop": 25440.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 13.333333333333334,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 13.333333333333334,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0649683,
    "mean_cost_per_repair_loop": 0.0216561,
    "p95_cost_per_repair_loop": 0.0288839,
    "cpsc": 0.0216561,
    "realized_cpsc": 0.0216561,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0649683,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0216561,
    "mean_spend_per_solved_task": 0.0216561,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.02348502666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0018289266666666655,
    "cost_delta_vs_gateway_ratio": -0.07787628656443221
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 41462.0,
    "tokens_per_success": 41462.0,
    "conditional_tokens_among_solved_loops": 41462.0,
    "mean_tokens_per_solved_task": 41462.0,
    "mean_input_tokens_per_repair_loop": 39293.0,
    "mean_output_tokens_per_repair_loop": 2169.0,
    "mean_cache_read_tokens_per_repair_loop": 28032.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 13.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 13.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.060388319999999995,
    "mean_cost_per_repair_loop": 0.02012944,
    "p95_cost_per_repair_loop": 0.0243633,
    "cpsc": 0.02012944,
    "realized_cpsc": 0.02012944,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.060388319999999995,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02012944,
    "mean_spend_per_solved_task": 0.02012944,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.020212863333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -8.342333333333692e-05,
    "cost_delta_vs_gateway_ratio": -0.004127239765964392
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 145066.66666666666,
    "tokens_per_success": 145066.66666666666,
    "conditional_tokens_among_solved_loops": 145066.66666666666,
    "mean_tokens_per_solved_task": 145066.66666666666,
    "mean_input_tokens_per_repair_loop": 139904.66666666666,
    "mean_output_tokens_per_repair_loop": 5162.0,
    "mean_cache_read_tokens_per_repair_loop": 108768.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 19.0,
    "p95_turns_per_repair_loop": 24.0,
    "mean_agent_steps_per_repair_loop": 19.0,
    "p95_agent_steps_per_repair_loop": 24.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.17226999999999998,
    "mean_cost_per_repair_loop": 0.057423333333333326,
    "p95_cost_per_repair_loop": 0.09492297999999999,
    "cpsc": 0.057423333333333326,
    "realized_cpsc": 0.057423333333333326,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.17226999999999998,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.057423333333333326,
    "mean_spend_per_solved_task": 0.057423333333333326,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06508045266666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.007657119333333341,
    "cost_delta_vs_gateway_ratio": -0.11765620888582441
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 33281.666666666664,
    "tokens_per_success": 33281.666666666664,
    "conditional_tokens_among_solved_loops": 33281.666666666664,
    "mean_tokens_per_solved_task": 33281.666666666664,
    "mean_input_tokens_per_repair_loop": 31149.666666666668,
    "mean_output_tokens_per_repair_loop": 2132.0,
    "mean_cache_read_tokens_per_repair_loop": 21498.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 13.0,
    "p95_turns_per_repair_loop": 19.0,
    "mean_agent_steps_per_repair_loop": 13.0,
    "p95_agent_steps_per_repair_loop": 19.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.053485620000000005,
    "mean_cost_per_repair_loop": 0.01782854,
    "p95_cost_per_repair_loop": 0.027592400000000003,
    "cpsc": 0.01782854,
    "realized_cpsc": 0.01782854,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.053485620000000005,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01782854,
    "mean_spend_per_solved_task": 0.01782854,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.019858816666666668,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.002030276666666666,
    "cost_delta_vs_gateway_ratio": -0.10223553098581735
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15609.666666666666,
    "tokens_per_success": 15609.666666666666,
    "conditional_tokens_among_solved_loops": 15609.666666666666,
    "mean_tokens_per_solved_task": 15609.666666666666,
    "mean_input_tokens_per_repair_loop": 14300.333333333334,
    "mean_output_tokens_per_repair_loop": 1309.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 8106.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.03114594,
    "mean_cost_per_repair_loop": 0.01038198,
    "p95_cost_per_repair_loop": 0.01196436,
    "cpsc": 0.01038198,
    "realized_cpsc": 0.01038198,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.03114594,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01038198,
    "mean_spend_per_solved_task": 0.01038198,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.011401843333333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0010198633333333349,
    "cost_delta_vs_gateway_ratio": -0.0894472326550708
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 55652.0,
    "tokens_per_success": 55652.0,
    "conditional_tokens_among_solved_loops": 55652.0,
    "mean_tokens_per_solved_task": 55652.0,
    "mean_input_tokens_per_repair_loop": 52600.333333333336,
    "mean_output_tokens_per_repair_loop": 3051.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 35888.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 12.666666666666666,
    "p95_turns_per_repair_loop": 17.0,
    "mean_agent_steps_per_repair_loop": 12.666666666666666,
    "p95_agent_steps_per_repair_loop": 17.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.08529348,
    "mean_cost_per_repair_loop": 0.02843116,
    "p95_cost_per_repair_loop": 0.04584324,
    "cpsc": 0.02843116,
    "realized_cpsc": 0.02843116,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.08529348,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02843116,
    "mean_spend_per_solved_task": 0.02843116,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.030728946666666663,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0022977866666666624,
    "cost_delta_vs_gateway_ratio": -0.07477596585369439
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 830712.0,
    "tokens_per_success": 830712.0,
    "conditional_tokens_among_solved_loops": 830712.0,
    "mean_tokens_per_solved_task": 830712.0,
    "mean_input_tokens_per_repair_loop": 799955.3333333334,
    "mean_output_tokens_per_repair_loop": 30756.666666666668,
    "mean_cache_read_tokens_per_repair_loop": 711546.6666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 31.0,
    "p95_turns_per_repair_loop": 53.0,
    "mean_agent_steps_per_repair_loop": 31.0,
    "p95_agent_steps_per_repair_loop": 53.0,
    "mean_verifier_submissions_per_repair_loop": 1.6666666666666667,
    "mean_verifier_submissions_to_success": 1.6666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.83940824,
    "mean_cost_per_repair_loop": 0.27980274666666666,
    "p95_cost_per_repair_loop": 0.48132194,
    "cpsc": 0.27980274666666666,
    "realized_cpsc": 0.27980274666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.83940824,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.27980274666666666,
    "mean_spend_per_solved_task": 0.27980274666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.3111348199999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.031332073333333245,
    "cost_delta_vs_gateway_ratio": -0.10070256145979821
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 101905.33333333333,
    "tokens_per_success": 101905.33333333333,
    "conditional_tokens_among_solved_loops": 101905.33333333333,
    "mean_tokens_per_solved_task": 101905.33333333333,
    "mean_input_tokens_per_repair_loop": 94819.66666666667,
    "mean_output_tokens_per_repair_loop": 7085.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 74778.66666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 13.666666666666666,
    "p95_turns_per_repair_loop": 27.0,
    "mean_agent_steps_per_repair_loop": 13.666666666666666,
    "p95_agent_steps_per_repair_loop": 27.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.15254092,
    "mean_cost_per_repair_loop": 0.05084697333333333,
    "p95_cost_per_repair_loop": 0.13547736,
    "cpsc": 0.05084697333333333,
    "realized_cpsc": 0.05084697333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.15254092,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.05084697333333333,
    "mean_spend_per_solved_task": 0.05084697333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.05419922133333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.003352248000000002,
    "cost_delta_vs_gateway_ratio": -0.061850482673601055
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 379033.0,
    "tokens_per_success": 379033.0,
    "conditional_tokens_among_solved_loops": 379033.0,
    "mean_tokens_per_solved_task": 379033.0,
    "mean_input_tokens_per_repair_loop": 359549.6666666667,
    "mean_output_tokens_per_repair_loop": 19483.333333333332,
    "mean_cache_read_tokens_per_repair_loop": 307616.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 28.666666666666668,
    "p95_turns_per_repair_loop": 31.0,
    "mean_agent_steps_per_repair_loop": 28.666666666666668,
    "p95_agent_steps_per_repair_loop": 31.0,
    "mean_verifier_submissions_per_repair_loop": 3.3333333333333335,
    "mean_verifier_submissions_to_success": 3.3333333333333335,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.45829494000000004,
    "mean_cost_per_repair_loop": 0.15276498000000002,
    "p95_cost_per_repair_loop": 0.22189722,
    "cpsc": 0.15276498000000002,
    "realized_cpsc": 0.15276498000000002,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.45829494000000004,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.15276498000000002,
    "mean_spend_per_solved_task": 0.15276498000000002,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.16824878333333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.015483803333333324,
    "cost_delta_vs_gateway_ratio": -0.09202921427762671
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 19193.0,
    "tokens_per_success": 19193.0,
    "conditional_tokens_among_solved_loops": 19193.0,
    "mean_tokens_per_solved_task": 19193.0,
    "mean_input_tokens_per_repair_loop": 17846.333333333332,
    "mean_output_tokens_per_repair_loop": 1346.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 11402.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.333333333333334,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 9.333333333333334,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.03357614,
    "mean_cost_per_repair_loop": 0.011192046666666665,
    "p95_cost_per_repair_loop": 0.01223562,
    "cpsc": 0.011192046666666665,
    "realized_cpsc": 0.011192046666666665,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.03357614,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.011192046666666665,
    "mean_spend_per_solved_task": 0.011192046666666665,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.011722723333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0005306766666666671,
    "cost_delta_vs_gateway_ratio": -0.04526906006198222
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23326.333333333332,
    "tokens_per_success": 23326.333333333332,
    "conditional_tokens_among_solved_loops": 23326.333333333332,
    "mean_tokens_per_solved_task": 23326.333333333332,
    "mean_input_tokens_per_repair_loop": 22228.0,
    "mean_output_tokens_per_repair_loop": 1098.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 12704.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.03839258,
    "mean_cost_per_repair_loop": 0.012797526666666668,
    "p95_cost_per_repair_loop": 0.01504064,
    "cpsc": 0.012797526666666668,
    "realized_cpsc": 0.012797526666666668,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.03839258,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.012797526666666668,
    "mean_spend_per_solved_task": 0.012797526666666668,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.013944110000000001,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0011465833333333328,
    "cost_delta_vs_gateway_ratio": -0.08222707174092378
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 16587.0,
    "tokens_per_success": 16587.0,
    "conditional_tokens_among_solved_loops": 16587.0,
    "mean_tokens_per_solved_task": 16587.0,
    "mean_input_tokens_per_repair_loop": 15395.333333333334,
    "mean_output_tokens_per_repair_loop": 1191.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 9696.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.02952822,
    "mean_cost_per_repair_loop": 0.00984274,
    "p95_cost_per_repair_loop": 0.012018,
    "cpsc": 0.00984274,
    "realized_cpsc": 0.00984274,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.02952822,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.00984274,
    "mean_spend_per_solved_task": 0.00984274,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.010563206666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.000720466666666667,
    "cost_delta_vs_gateway_ratio": -0.06820529876975491
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 167017.0,
    "tokens_per_success": 167017.0,
    "conditional_tokens_among_solved_loops": 167017.0,
    "mean_tokens_per_solved_task": 167017.0,
    "mean_input_tokens_per_repair_loop": 156186.0,
    "mean_output_tokens_per_repair_loop": 10831.0,
    "mean_cache_read_tokens_per_repair_loop": 125717.33333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 16.0,
    "p95_turns_per_repair_loop": 23.0,
    "mean_agent_steps_per_repair_loop": 16.0,
    "p95_agent_steps_per_repair_loop": 23.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.23793873999999998,
    "mean_cost_per_repair_loop": 0.07931291333333333,
    "p95_cost_per_repair_loop": 0.09628462,
    "cpsc": 0.07931291333333333,
    "realized_cpsc": 0.07931291333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.23793873999999998,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07931291333333333,
    "mean_spend_per_solved_task": 0.07931291333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08906569666666665,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.009752783333333329,
    "cost_delta_vs_gateway_ratio": -0.10950100541888383
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 14935.666666666666,
    "tokens_per_success": 14935.666666666666,
    "conditional_tokens_among_solved_loops": 14935.666666666666,
    "mean_tokens_per_solved_task": 14935.666666666666,
    "mean_input_tokens_per_repair_loop": 13863.333333333334,
    "mean_output_tokens_per_repair_loop": 1072.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 9488.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.025242340000000002,
    "mean_cost_per_repair_loop": 0.008414113333333334,
    "p95_cost_per_repair_loop": 0.01123266,
    "cpsc": 0.008414113333333334,
    "realized_cpsc": 0.008414113333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.025242340000000002,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.008414113333333334,
    "mean_spend_per_solved_task": 0.008414113333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.008399263333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": 1.4850000000001437e-05,
    "cost_delta_vs_gateway_ratio": 0.0017680121947204223
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 88678.66666666667,
    "tokens_per_success": 88678.66666666667,
    "conditional_tokens_among_solved_loops": 88678.66666666667,
    "mean_tokens_per_solved_task": 88678.66666666667,
    "mean_input_tokens_per_repair_loop": 84336.33333333333,
    "mean_output_tokens_per_repair_loop": 4342.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 63765.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 16.333333333333332,
    "p95_turns_per_repair_loop": 27.0,
    "mean_agent_steps_per_repair_loop": 16.333333333333332,
    "p95_agent_steps_per_repair_loop": 27.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.11995652,
    "mean_cost_per_repair_loop": 0.03998550666666666,
    "p95_cost_per_repair_loop": 0.07513818,
    "cpsc": 0.03998550666666666,
    "realized_cpsc": 0.03998550666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.11995652,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.03998550666666666,
    "mean_spend_per_solved_task": 0.03998550666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04378443333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.0037989266666666696,
    "cost_delta_vs_gateway_ratio": -0.08676432187086285
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 3429863.6666666665,
    "tokens_per_success": 3429863.6666666665,
    "conditional_tokens_among_solved_loops": 3429863.6666666665,
    "mean_tokens_per_solved_task": 3429863.6666666665,
    "mean_input_tokens_per_repair_loop": 3351768.0,
    "mean_output_tokens_per_repair_loop": 78095.66666666667,
    "mean_cache_read_tokens_per_repair_loop": 3210240.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 56136.333333333336,
    "mean_turns_per_repair_loop": 63.666666666666664,
    "p95_turns_per_repair_loop": 67.0,
    "mean_agent_steps_per_repair_loop": 63.666666666666664,
    "p95_agent_steps_per_repair_loop": 67.0,
    "mean_verifier_submissions_per_repair_loop": 5.666666666666667,
    "mean_verifier_submissions_to_success": 5.666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 7.677273,
    "mean_cost_per_repair_loop": 2.559091,
    "p95_cost_per_repair_loop": 2.9381886,
    "cpsc": 2.559091,
    "realized_cpsc": 2.559091,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 7.677273,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 2.559091,
    "mean_spend_per_solved_task": 2.559091,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 2.3826468000000003,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.1764441999999994,
    "cost_delta_vs_gateway_ratio": 0.07405386312398438
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 58461.0,
    "tokens_per_success": 58461.0,
    "conditional_tokens_among_solved_loops": 58461.0,
    "mean_tokens_per_solved_task": 58461.0,
    "mean_input_tokens_per_repair_loop": 52196.0,
    "mean_output_tokens_per_repair_loop": 6265.0,
    "mean_cache_read_tokens_per_repair_loop": 41472.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 3224.6666666666665,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 12.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 12.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.4157658,
    "mean_cost_per_repair_loop": 0.1385886,
    "p95_cost_per_repair_loop": 0.1743294,
    "cpsc": 0.1385886,
    "realized_cpsc": 0.1385886,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.4157658,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1385886,
    "mean_spend_per_solved_task": 0.1385886,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1385886,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 86434.66666666667,
    "tokens_per_success": 86434.66666666667,
    "conditional_tokens_among_solved_loops": 86434.66666666667,
    "mean_tokens_per_solved_task": 86434.66666666667,
    "mean_input_tokens_per_repair_loop": 71393.33333333333,
    "mean_output_tokens_per_repair_loop": 15041.333333333334,
    "mean_cache_read_tokens_per_repair_loop": 57002.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 9789.0,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.8576784,
    "mean_cost_per_repair_loop": 0.2858928,
    "p95_cost_per_repair_loop": 0.30614159999999996,
    "cpsc": 0.2858928,
    "realized_cpsc": 0.2858928,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.8576784,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.2858928,
    "mean_spend_per_solved_task": 0.2858928,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.15409640000000002,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.13179639999999995,
    "cost_delta_vs_gateway_ratio": 0.8552853927801035
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 100917.66666666667,
    "tokens_per_success": 100917.66666666667,
    "conditional_tokens_among_solved_loops": 100917.66666666667,
    "mean_tokens_per_solved_task": 100917.66666666667,
    "mean_input_tokens_per_repair_loop": 92004.33333333333,
    "mean_output_tokens_per_repair_loop": 8913.333333333334,
    "mean_cache_read_tokens_per_repair_loop": 78250.66666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 4640.666666666667,
    "mean_turns_per_repair_loop": 13.333333333333334,
    "p95_turns_per_repair_loop": 15.0,
    "mean_agent_steps_per_repair_loop": 13.333333333333334,
    "p95_agent_steps_per_repair_loop": 15.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5953086000000001,
    "mean_cost_per_repair_loop": 0.19843620000000003,
    "p95_cost_per_repair_loop": 0.2461182,
    "cpsc": 0.19843620000000003,
    "realized_cpsc": 0.19843620000000003,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5953086000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.19843620000000003,
    "mean_spend_per_solved_task": 0.19843620000000003,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.19843619999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": 1.8649537813247728e-16
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 37993.666666666664,
    "tokens_per_success": 37993.666666666664,
    "conditional_tokens_among_solved_loops": 37993.666666666664,
    "mean_tokens_per_solved_task": 37993.666666666664,
    "mean_input_tokens_per_repair_loop": 34303.666666666664,
    "mean_output_tokens_per_repair_loop": 3690.0,
    "mean_cache_read_tokens_per_repair_loop": 26368.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1722.3333333333333,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2612022,
    "mean_cost_per_repair_loop": 0.0870674,
    "p95_cost_per_repair_loop": 0.0958146,
    "cpsc": 0.0870674,
    "realized_cpsc": 0.0870674,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2612022,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0870674,
    "mean_spend_per_solved_task": 0.0870674,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0870674,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 169825.0,
    "tokens_per_success": 169825.0,
    "conditional_tokens_among_solved_loops": 169825.0,
    "mean_tokens_per_solved_task": 169825.0,
    "mean_input_tokens_per_repair_loop": 159890.0,
    "mean_output_tokens_per_repair_loop": 9935.0,
    "mean_cache_read_tokens_per_repair_loop": 140373.33333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 4626.333333333333,
    "mean_turns_per_repair_loop": 15.666666666666666,
    "p95_turns_per_repair_loop": 17.0,
    "mean_agent_steps_per_repair_loop": 15.666666666666666,
    "p95_agent_steps_per_repair_loop": 17.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.749061,
    "mean_cost_per_repair_loop": 0.249687,
    "p95_cost_per_repair_loop": 0.3301944,
    "cpsc": 0.249687,
    "realized_cpsc": 0.249687,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.749061,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.249687,
    "mean_spend_per_solved_task": 0.249687,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.24968700000000002,
    "mean_cost_delta_vs_gateway_per_repair_loop": -3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": -1.4821530217501063e-16
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 51760.333333333336,
    "tokens_per_success": 51760.333333333336,
    "conditional_tokens_among_solved_loops": 51760.333333333336,
    "mean_tokens_per_solved_task": 51760.333333333336,
    "mean_input_tokens_per_repair_loop": 45111.0,
    "mean_output_tokens_per_repair_loop": 6649.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 35754.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 4434.666666666667,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.41560620000000004,
    "mean_cost_per_repair_loop": 0.1385354,
    "p95_cost_per_repair_loop": 0.1577838,
    "cpsc": 0.1385354,
    "realized_cpsc": 0.1385354,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.41560620000000004,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1385354,
    "mean_spend_per_solved_task": 0.1385354,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1385354,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 46700.333333333336,
    "tokens_per_success": 46700.333333333336,
    "conditional_tokens_among_solved_loops": 46700.333333333336,
    "mean_tokens_per_solved_task": 46700.333333333336,
    "mean_input_tokens_per_repair_loop": 42493.333333333336,
    "mean_output_tokens_per_repair_loop": 4207.0,
    "mean_cache_read_tokens_per_repair_loop": 33877.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1925.3333333333333,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.29734859999999996,
    "mean_cost_per_repair_loop": 0.09911619999999999,
    "p95_cost_per_repair_loop": 0.1093488,
    "cpsc": 0.09911619999999999,
    "realized_cpsc": 0.09911619999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.29734859999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09911619999999999,
    "mean_spend_per_solved_task": 0.09911619999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0991162,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.866871114619602e-16
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 56222.0,
    "tokens_per_success": 56222.0,
    "conditional_tokens_among_solved_loops": 56222.0,
    "mean_tokens_per_solved_task": 56222.0,
    "mean_input_tokens_per_repair_loop": 51670.666666666664,
    "mean_output_tokens_per_repair_loop": 4551.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 41557.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1416.6666666666667,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.33323159999999996,
    "mean_cost_per_repair_loop": 0.11107719999999999,
    "p95_cost_per_repair_loop": 0.12440939999999999,
    "cpsc": 0.11107719999999999,
    "realized_cpsc": 0.11107719999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.33323159999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11107719999999999,
    "mean_spend_per_solved_task": 0.11107719999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11107719999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 152600.0,
    "tokens_per_success": 152600.0,
    "conditional_tokens_among_solved_loops": 152600.0,
    "mean_tokens_per_solved_task": 152600.0,
    "mean_input_tokens_per_repair_loop": 136560.66666666666,
    "mean_output_tokens_per_repair_loop": 16039.333333333334,
    "mean_cache_read_tokens_per_repair_loop": 119381.33333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 10337.333333333334,
    "mean_turns_per_repair_loop": 12.333333333333334,
    "p95_turns_per_repair_loop": 17.0,
    "mean_agent_steps_per_repair_loop": 12.333333333333334,
    "p95_agent_steps_per_repair_loop": 17.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.9838272,
    "mean_cost_per_repair_loop": 0.3279424,
    "p95_cost_per_repair_loop": 0.4034988,
    "cpsc": 0.3279424,
    "realized_cpsc": 0.3279424,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.9838272,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.3279424,
    "mean_spend_per_solved_task": 0.3279424,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2423864,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.08555600000000001,
    "cost_delta_vs_gateway_ratio": 0.3529735991788318
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 54769.0,
    "tokens_per_success": 54769.0,
    "conditional_tokens_among_solved_loops": 54769.0,
    "mean_tokens_per_solved_task": 54769.0,
    "mean_input_tokens_per_repair_loop": 47739.333333333336,
    "mean_output_tokens_per_repair_loop": 7029.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 37120.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 3693.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.44531699999999996,
    "mean_cost_per_repair_loop": 0.148439,
    "p95_cost_per_repair_loop": 0.162459,
    "cpsc": 0.148439,
    "realized_cpsc": 0.148439,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.44531699999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.148439,
    "mean_spend_per_solved_task": 0.148439,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14843900000000002,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.2465536063356625e-16
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 117464.0,
    "tokens_per_success": 117464.0,
    "conditional_tokens_among_solved_loops": 117464.0,
    "mean_tokens_per_solved_task": 117464.0,
    "mean_input_tokens_per_repair_loop": 108571.66666666667,
    "mean_output_tokens_per_repair_loop": 8892.333333333334,
    "mean_cache_read_tokens_per_repair_loop": 91904.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 5424.333333333333,
    "mean_turns_per_repair_loop": 12.333333333333334,
    "p95_turns_per_repair_loop": 19.0,
    "mean_agent_steps_per_repair_loop": 12.333333333333334,
    "p95_agent_steps_per_repair_loop": 19.0,
    "mean_verifier_submissions_per_repair_loop": 1.6666666666666667,
    "mean_verifier_submissions_to_success": 1.6666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.6328776,
    "mean_cost_per_repair_loop": 0.2109592,
    "p95_cost_per_repair_loop": 0.341085,
    "cpsc": 0.2109592,
    "realized_cpsc": 0.2109592,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.6328776,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.2109592,
    "mean_spend_per_solved_task": 0.2109592,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2109592,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 27961.0,
    "tokens_per_success": 27961.0,
    "conditional_tokens_among_solved_loops": 27961.0,
    "mean_tokens_per_solved_task": 27961.0,
    "mean_input_tokens_per_repair_loop": 24565.0,
    "mean_output_tokens_per_repair_loop": 3396.0,
    "mean_cache_read_tokens_per_repair_loop": 18261.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1323.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2259882,
    "mean_cost_per_repair_loop": 0.0753294,
    "p95_cost_per_repair_loop": 0.0833382,
    "cpsc": 0.0753294,
    "realized_cpsc": 0.0753294,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2259882,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0753294,
    "mean_spend_per_solved_task": 0.0753294,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07532939999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 9.25185853854297e-18,
    "cost_delta_vs_gateway_ratio": 1.228186941425655e-16
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 52830.666666666664,
    "tokens_per_success": 52830.666666666664,
    "conditional_tokens_among_solved_loops": 52830.666666666664,
    "mean_tokens_per_solved_task": 52830.666666666664,
    "mean_input_tokens_per_repair_loop": 48875.0,
    "mean_output_tokens_per_repair_loop": 3955.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 39936.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1439.3333333333333,
    "mean_turns_per_repair_loop": 12.0,
    "p95_turns_per_repair_loop": 13.0,
    "mean_agent_steps_per_repair_loop": 12.0,
    "p95_agent_steps_per_repair_loop": 13.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2943984,
    "mean_cost_per_repair_loop": 0.0981328,
    "p95_cost_per_repair_loop": 0.10942560000000001,
    "cpsc": 0.0981328,
    "realized_cpsc": 0.0981328,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2943984,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0981328,
    "mean_spend_per_solved_task": 0.0981328,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0981328,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 43473.0,
    "tokens_per_success": 43473.0,
    "conditional_tokens_among_solved_loops": 43473.0,
    "mean_tokens_per_solved_task": 43473.0,
    "mean_input_tokens_per_repair_loop": 39235.0,
    "mean_output_tokens_per_repair_loop": 4238.0,
    "mean_cache_read_tokens_per_repair_loop": 30890.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1598.6666666666667,
    "mean_turns_per_repair_loop": 10.333333333333334,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 10.333333333333334,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2936106,
    "mean_cost_per_repair_loop": 0.0978702,
    "p95_cost_per_repair_loop": 0.1034688,
    "cpsc": 0.0978702,
    "realized_cpsc": 0.0978702,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2936106,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0978702,
    "mean_spend_per_solved_task": 0.0978702,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0978702,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 91095.33333333333,
    "tokens_per_success": 91095.33333333333,
    "conditional_tokens_among_solved_loops": 91095.33333333333,
    "mean_tokens_per_solved_task": 91095.33333333333,
    "mean_input_tokens_per_repair_loop": 74449.0,
    "mean_output_tokens_per_repair_loop": 16646.333333333332,
    "mean_cache_read_tokens_per_repair_loop": 58965.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 11499.666666666666,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.9415068,
    "mean_cost_per_repair_loop": 0.3138356,
    "p95_cost_per_repair_loop": 0.35379,
    "cpsc": 0.3138356,
    "realized_cpsc": 0.3138356,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.9415068,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.3138356,
    "mean_spend_per_solved_task": 0.3138356,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1856208,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.1282148,
    "cost_delta_vs_gateway_ratio": 0.6907350900330136
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 33844.666666666664,
    "tokens_per_success": 33844.666666666664,
    "conditional_tokens_among_solved_loops": 33844.666666666664,
    "mean_tokens_per_solved_task": 33844.666666666664,
    "mean_input_tokens_per_repair_loop": 30011.0,
    "mean_output_tokens_per_repair_loop": 3833.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 23210.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1951.0,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2546076,
    "mean_cost_per_repair_loop": 0.08486919999999999,
    "p95_cost_per_repair_loop": 0.09529439999999999,
    "cpsc": 0.08486919999999999,
    "realized_cpsc": 0.08486919999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2546076,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.08486919999999999,
    "mean_spend_per_solved_task": 0.08486919999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08486919999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 1753461.0,
    "tokens_per_success": 1753461.0,
    "conditional_tokens_among_solved_loops": 1753461.0,
    "mean_tokens_per_solved_task": 1753461.0,
    "mean_input_tokens_per_repair_loop": 1699080.6666666667,
    "mean_output_tokens_per_repair_loop": 54380.333333333336,
    "mean_cache_read_tokens_per_repair_loop": 1623210.6666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 37163.333333333336,
    "mean_turns_per_repair_loop": 46.333333333333336,
    "p95_turns_per_repair_loop": 62.0,
    "mean_agent_steps_per_repair_loop": 46.333333333333336,
    "p95_agent_steps_per_repair_loop": 62.0,
    "mean_verifier_submissions_per_repair_loop": 5.333333333333333,
    "mean_verifier_submissions_to_success": 5.333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 4.590834599999999,
    "mean_cost_per_repair_loop": 1.5302781999999997,
    "p95_cost_per_repair_loop": 2.0286617999999996,
    "cpsc": 1.5302781999999997,
    "realized_cpsc": 1.5302781999999997,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 4.590834599999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 1.5302781999999997,
    "mean_spend_per_solved_task": 1.5302781999999997,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 1.5302782000000004,
    "mean_cost_delta_vs_gateway_per_repair_loop": -5.921189464667501e-16,
    "cost_delta_vs_gateway_ratio": -3.8693549085829625e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23850.666666666668,
    "tokens_per_success": 23850.666666666668,
    "conditional_tokens_among_solved_loops": 23850.666666666668,
    "mean_tokens_per_solved_task": 23850.666666666668,
    "mean_input_tokens_per_repair_loop": 21098.0,
    "mean_output_tokens_per_repair_loop": 2752.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 11605.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.40753799999999996,
    "mean_cost_per_repair_loop": 0.135846,
    "p95_cost_per_repair_loop": 0.148745,
    "cpsc": 0.135846,
    "realized_cpsc": 0.135846,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.40753799999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.135846,
    "mean_spend_per_solved_task": 0.135846,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.135846,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.3621098212009144e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15557.333333333334,
    "tokens_per_success": 15557.333333333334,
    "conditional_tokens_among_solved_loops": 15557.333333333334,
    "mean_tokens_per_solved_task": 15557.333333333334,
    "mean_input_tokens_per_repair_loop": 13823.666666666666,
    "mean_output_tokens_per_repair_loop": 1733.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 7850.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.257401,
    "mean_cost_per_repair_loop": 0.08580033333333333,
    "p95_cost_per_repair_loop": 0.088106,
    "cpsc": 0.08580033333333333,
    "realized_cpsc": 0.08580033333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.257401,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.08580033333333333,
    "mean_spend_per_solved_task": 0.08580033333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08580033333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 27685.333333333332,
    "tokens_per_success": 27685.333333333332,
    "conditional_tokens_among_solved_loops": 27685.333333333332,
    "mean_tokens_per_solved_task": 27685.333333333332,
    "mean_input_tokens_per_repair_loop": 24853.333333333332,
    "mean_output_tokens_per_repair_loop": 2832.0,
    "mean_cache_read_tokens_per_repair_loop": 13141.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.450272,
    "mean_cost_per_repair_loop": 0.15009066666666668,
    "p95_cost_per_repair_loop": 0.181867,
    "cpsc": 0.15009066666666668,
    "realized_cpsc": 0.15009066666666668,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.450272,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.15009066666666668,
    "mean_spend_per_solved_task": 0.15009066666666668,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.13666199999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.01342866666666668,
    "cost_delta_vs_gateway_ratio": 0.09826189186947859
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 18959.333333333332,
    "tokens_per_success": 18959.333333333332,
    "conditional_tokens_among_solved_loops": 18959.333333333332,
    "mean_tokens_per_solved_task": 18959.333333333332,
    "mean_input_tokens_per_repair_loop": 16889.333333333332,
    "mean_output_tokens_per_repair_loop": 2070.0,
    "mean_cache_read_tokens_per_repair_loop": 6144.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.356696,
    "mean_cost_per_repair_loop": 0.11889866666666667,
    "p95_cost_per_repair_loop": 0.12845,
    "cpsc": 0.11889866666666667,
    "realized_cpsc": 0.11889866666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.356696,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11889866666666667,
    "mean_spend_per_solved_task": 0.11889866666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11889866666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 35617.0,
    "tokens_per_success": 35617.0,
    "conditional_tokens_among_solved_loops": 35617.0,
    "mean_tokens_per_solved_task": 35617.0,
    "mean_input_tokens_per_repair_loop": 33720.333333333336,
    "mean_output_tokens_per_repair_loop": 1896.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 18944.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.0,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 9.0,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.420761,
    "mean_cost_per_repair_loop": 0.14025366666666667,
    "p95_cost_per_repair_loop": 0.176384,
    "cpsc": 0.14025366666666667,
    "realized_cpsc": 0.14025366666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.420761,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14025366666666667,
    "mean_spend_per_solved_task": 0.14025366666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14025366666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 46424.666666666664,
    "tokens_per_success": 46424.666666666664,
    "conditional_tokens_among_solved_loops": 46424.666666666664,
    "mean_tokens_per_solved_task": 46424.666666666664,
    "mean_input_tokens_per_repair_loop": 43314.333333333336,
    "mean_output_tokens_per_repair_loop": 3110.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 20821.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.648557,
    "mean_cost_per_repair_loop": 0.2161856666666667,
    "p95_cost_per_repair_loop": 0.237661,
    "cpsc": 0.2161856666666667,
    "realized_cpsc": 0.2161856666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.648557,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.2161856666666667,
    "mean_spend_per_solved_task": 0.2161856666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.21618566666666664,
    "mean_cost_delta_vs_gateway_per_repair_loop": 3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": 1.7118356977492443e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 20054.666666666668,
    "tokens_per_success": 20054.666666666668,
    "conditional_tokens_among_solved_loops": 20054.666666666668,
    "mean_tokens_per_solved_task": 20054.666666666668,
    "mean_input_tokens_per_repair_loop": 18354.0,
    "mean_output_tokens_per_repair_loop": 1700.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 7509.333333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.326994,
    "mean_cost_per_repair_loop": 0.108998,
    "p95_cost_per_repair_loop": 0.155673,
    "cpsc": 0.108998,
    "realized_cpsc": 0.108998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.326994,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.108998,
    "mean_spend_per_solved_task": 0.108998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.108998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 12193.0,
    "tokens_per_success": 12193.0,
    "conditional_tokens_among_solved_loops": 12193.0,
    "mean_tokens_per_solved_task": 12193.0,
    "mean_input_tokens_per_repair_loop": 10801.0,
    "mean_output_tokens_per_repair_loop": 1392.0,
    "mean_cache_read_tokens_per_repair_loop": 4608.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.22508699999999998,
    "mean_cost_per_repair_loop": 0.075029,
    "p95_cost_per_repair_loop": 0.081439,
    "cpsc": 0.075029,
    "realized_cpsc": 0.075029,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.22508699999999998,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.075029,
    "mean_spend_per_solved_task": 0.075029,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.075029,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 32486.333333333332,
    "tokens_per_success": 32486.333333333332,
    "conditional_tokens_among_solved_loops": 32486.333333333332,
    "mean_tokens_per_solved_task": 32486.333333333332,
    "mean_input_tokens_per_repair_loop": 30474.333333333332,
    "mean_output_tokens_per_repair_loop": 2012.0,
    "mean_cache_read_tokens_per_repair_loop": 17066.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 9.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.407795,
    "mean_cost_per_repair_loop": 0.13593166666666667,
    "p95_cost_per_repair_loop": 0.158057,
    "cpsc": 0.13593166666666667,
    "realized_cpsc": 0.13593166666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.407795,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.13593166666666667,
    "mean_spend_per_solved_task": 0.13593166666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.13593166666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 36862.666666666664,
    "tokens_per_success": 36862.666666666664,
    "conditional_tokens_among_solved_loops": 36862.666666666664,
    "mean_tokens_per_solved_task": 36862.666666666664,
    "mean_input_tokens_per_repair_loop": 33076.666666666664,
    "mean_output_tokens_per_repair_loop": 3786.0,
    "mean_cache_read_tokens_per_repair_loop": 14336.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.643354,
    "mean_cost_per_repair_loop": 0.21445133333333333,
    "p95_cost_per_repair_loop": 0.244623,
    "cpsc": 0.21445133333333333,
    "realized_cpsc": 0.21445133333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.643354,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.21445133333333333,
    "mean_spend_per_solved_task": 0.21445133333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.21445133333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 19796.333333333332,
    "tokens_per_success": 19796.333333333332,
    "conditional_tokens_among_solved_loops": 19796.333333333332,
    "mean_tokens_per_solved_task": 19796.333333333332,
    "mean_input_tokens_per_repair_loop": 17508.333333333332,
    "mean_output_tokens_per_repair_loop": 2288.0,
    "mean_cache_read_tokens_per_repair_loop": 10240.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.33030499999999996,
    "mean_cost_per_repair_loop": 0.11010166666666665,
    "p95_cost_per_repair_loop": 0.121973,
    "cpsc": 0.11010166666666665,
    "realized_cpsc": 0.11010166666666665,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.33030499999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11010166666666665,
    "mean_spend_per_solved_task": 0.11010166666666665,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11010166666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.6806028134983674e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 42106.666666666664,
    "tokens_per_success": 42106.666666666664,
    "conditional_tokens_among_solved_loops": 42106.666666666664,
    "mean_tokens_per_solved_task": 42106.666666666664,
    "mean_input_tokens_per_repair_loop": 39099.333333333336,
    "mean_output_tokens_per_repair_loop": 3007.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 25088.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.0,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 9.0,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.518462,
    "mean_cost_per_repair_loop": 0.17282066666666665,
    "p95_cost_per_repair_loop": 0.223198,
    "cpsc": 0.17282066666666665,
    "realized_cpsc": 0.17282066666666665,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.518462,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.17282066666666665,
    "mean_spend_per_solved_task": 0.17282066666666665,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1728206666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": -2.141377814816045e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 14259.666666666666,
    "tokens_per_success": 14259.666666666666,
    "conditional_tokens_among_solved_loops": 14259.666666666666,
    "mean_tokens_per_solved_task": 14259.666666666666,
    "mean_input_tokens_per_repair_loop": 12851.0,
    "mean_output_tokens_per_repair_loop": 1408.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 5802.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.241209,
    "mean_cost_per_repair_loop": 0.080403,
    "p95_cost_per_repair_loop": 0.084425,
    "cpsc": 0.080403,
    "realized_cpsc": 0.080403,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.241209,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.080403,
    "mean_spend_per_solved_task": 0.080403,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.080403,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 25159.0,
    "tokens_per_success": 25159.0,
    "conditional_tokens_among_solved_loops": 25159.0,
    "mean_tokens_per_solved_task": 25159.0,
    "mean_input_tokens_per_repair_loop": 22825.666666666668,
    "mean_output_tokens_per_repair_loop": 2333.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 12288.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.386497,
    "mean_cost_per_repair_loop": 0.12883233333333333,
    "p95_cost_per_repair_loop": 0.147557,
    "cpsc": 0.12883233333333333,
    "realized_cpsc": 0.12883233333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.386497,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12883233333333333,
    "mean_spend_per_solved_task": 0.12883233333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12883233333333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.4362634439920057e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 27451.0,
    "tokens_per_success": 27451.0,
    "conditional_tokens_among_solved_loops": 27451.0,
    "mean_tokens_per_solved_task": 27451.0,
    "mean_input_tokens_per_repair_loop": 25670.333333333332,
    "mean_output_tokens_per_repair_loop": 1780.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 12288.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.379427,
    "mean_cost_per_repair_loop": 0.12647566666666668,
    "p95_cost_per_repair_loop": 0.153674,
    "cpsc": 0.12647566666666668,
    "realized_cpsc": 0.12647566666666668,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.379427,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12647566666666668,
    "mean_spend_per_solved_task": 0.12647566666666668,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12647566666666668,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 35764.0,
    "tokens_per_success": 35764.0,
    "conditional_tokens_among_solved_loops": 35764.0,
    "mean_tokens_per_solved_task": 35764.0,
    "mean_input_tokens_per_repair_loop": 32165.0,
    "mean_output_tokens_per_repair_loop": 3599.0,
    "mean_cache_read_tokens_per_repair_loop": 18261.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.559857,
    "mean_cost_per_repair_loop": 0.186619,
    "p95_cost_per_repair_loop": 0.203127,
    "cpsc": 0.186619,
    "realized_cpsc": 0.186619,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.559857,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.186619,
    "mean_spend_per_solved_task": 0.186619,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.186619,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15459.666666666666,
    "tokens_per_success": 15459.666666666666,
    "conditional_tokens_among_solved_loops": 15459.666666666666,
    "mean_tokens_per_solved_task": 15459.666666666666,
    "mean_input_tokens_per_repair_loop": 13804.666666666666,
    "mean_output_tokens_per_repair_loop": 1655.0,
    "mean_cache_read_tokens_per_repair_loop": 6314.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.270772,
    "mean_cost_per_repair_loop": 0.09025733333333334,
    "p95_cost_per_repair_loop": 0.101749,
    "cpsc": 0.09025733333333334,
    "realized_cpsc": 0.09025733333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.270772,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09025733333333334,
    "mean_spend_per_solved_task": 0.09025733333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.09025733333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 18671.333333333332,
    "tokens_per_success": 18671.333333333332,
    "conditional_tokens_among_solved_loops": 18671.333333333332,
    "mean_tokens_per_solved_task": 18671.333333333332,
    "mean_input_tokens_per_repair_loop": 16474.333333333332,
    "mean_output_tokens_per_repair_loop": 2197.0,
    "mean_cache_read_tokens_per_repair_loop": 7338.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.345773,
    "mean_cost_per_repair_loop": 0.11525766666666666,
    "p95_cost_per_repair_loop": 0.126754,
    "cpsc": 0.11525766666666666,
    "realized_cpsc": 0.11525766666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.345773,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11525766666666666,
    "mean_spend_per_solved_task": 0.11525766666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11525766666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 30123.0,
    "tokens_per_success": 30123.0,
    "conditional_tokens_among_solved_loops": 30123.0,
    "mean_tokens_per_solved_task": 30123.0,
    "mean_input_tokens_per_repair_loop": 26701.333333333332,
    "mean_output_tokens_per_repair_loop": 3421.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 14165.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.517238,
    "mean_cost_per_repair_loop": 0.17241266666666666,
    "p95_cost_per_repair_loop": 0.199615,
    "cpsc": 0.17241266666666666,
    "realized_cpsc": 0.17241266666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.517238,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.17241266666666666,
    "mean_spend_per_solved_task": 0.17241266666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.17241266666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 24259.666666666668,
    "tokens_per_success": 24259.666666666668,
    "conditional_tokens_among_solved_loops": 24259.666666666668,
    "mean_tokens_per_solved_task": 24259.666666666668,
    "mean_input_tokens_per_repair_loop": 21615.0,
    "mean_output_tokens_per_repair_loop": 2644.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 8874.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.44243699999999997,
    "mean_cost_per_repair_loop": 0.147479,
    "p95_cost_per_repair_loop": 0.210085,
    "cpsc": 0.147479,
    "realized_cpsc": 0.147479,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.44243699999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.147479,
    "mean_spend_per_solved_task": 0.147479,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.147479,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.2546679240492506e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 33097.333333333336,
    "tokens_per_success": 33097.333333333336,
    "conditional_tokens_among_solved_loops": 33097.333333333336,
    "mean_tokens_per_solved_task": 33097.333333333336,
    "mean_input_tokens_per_repair_loop": 28616.666666666668,
    "mean_output_tokens_per_repair_loop": 4480.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 14848.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.632062,
    "mean_cost_per_repair_loop": 0.21068733333333334,
    "p95_cost_per_repair_loop": 0.275736,
    "cpsc": 0.21068733333333334,
    "realized_cpsc": 0.21068733333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.632062,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.21068733333333334,
    "mean_spend_per_solved_task": 0.21068733333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2106873333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": 1.7565096851656272e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 27308.666666666668,
    "tokens_per_success": 27308.666666666668,
    "conditional_tokens_among_solved_loops": 27308.666666666668,
    "mean_tokens_per_solved_task": 27308.666666666668,
    "mean_input_tokens_per_repair_loop": 24731.333333333332,
    "mean_output_tokens_per_repair_loop": 2577.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 13482.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 9.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.420914,
    "mean_cost_per_repair_loop": 0.14030466666666666,
    "p95_cost_per_repair_loop": 0.148851,
    "cpsc": 0.14030466666666666,
    "realized_cpsc": 0.14030466666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.420914,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14030466666666666,
    "mean_spend_per_solved_task": 0.14030466666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14030466666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 54458.333333333336,
    "tokens_per_success": 54458.333333333336,
    "conditional_tokens_among_solved_loops": 54458.333333333336,
    "mean_tokens_per_solved_task": 54458.333333333336,
    "mean_input_tokens_per_repair_loop": 51852.0,
    "mean_output_tokens_per_repair_loop": 2606.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 33962.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.333333333333334,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 10.333333333333334,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5538540000000001,
    "mean_cost_per_repair_loop": 0.18461800000000003,
    "p95_cost_per_repair_loop": 0.231601,
    "cpsc": 0.18461800000000003,
    "realized_cpsc": 0.18461800000000003,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5538540000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.18461800000000003,
    "mean_spend_per_solved_task": 0.18461800000000003,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.18461800000000003,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 87684.0,
    "tokens_per_success": 87684.0,
    "conditional_tokens_among_solved_loops": 87684.0,
    "mean_tokens_per_solved_task": 87684.0,
    "mean_input_tokens_per_repair_loop": 82156.66666666667,
    "mean_output_tokens_per_repair_loop": 5527.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 49664.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 12.0,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 12.0,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.0593460000000001,
    "mean_cost_per_repair_loop": 0.3531153333333334,
    "p95_cost_per_repair_loop": 0.390045,
    "cpsc": 0.3531153333333334,
    "realized_cpsc": 0.3531153333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.0593460000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.3531153333333334,
    "mean_spend_per_solved_task": 0.3531153333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.3531153333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17779.333333333332,
    "tokens_per_success": 17779.333333333332,
    "conditional_tokens_among_solved_loops": 17779.333333333332,
    "mean_tokens_per_solved_task": 17779.333333333332,
    "mean_input_tokens_per_repair_loop": 15529.666666666666,
    "mean_output_tokens_per_repair_loop": 2249.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 7509.333333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.334039,
    "mean_cost_per_repair_loop": 0.11134633333333332,
    "p95_cost_per_repair_loop": 0.121528,
    "cpsc": 0.11134633333333332,
    "realized_cpsc": 0.11134633333333332,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.334039,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11134633333333332,
    "mean_spend_per_solved_task": 0.11134633333333332,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11134633333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 26359.333333333332,
    "tokens_per_success": 26359.333333333332,
    "conditional_tokens_among_solved_loops": 26359.333333333332,
    "mean_tokens_per_solved_task": 26359.333333333332,
    "mean_input_tokens_per_repair_loop": 24024.666666666668,
    "mean_output_tokens_per_repair_loop": 2334.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 10752.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.425338,
    "mean_cost_per_repair_loop": 0.14177933333333334,
    "p95_cost_per_repair_loop": 0.221797,
    "cpsc": 0.14177933333333334,
    "realized_cpsc": 0.14177933333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.425338,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14177933333333334,
    "mean_spend_per_solved_task": 0.14177933333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14177933333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 32885.0,
    "tokens_per_success": 32885.0,
    "conditional_tokens_among_solved_loops": 32885.0,
    "mean_tokens_per_solved_task": 32885.0,
    "mean_input_tokens_per_repair_loop": 30368.333333333332,
    "mean_output_tokens_per_repair_loop": 2516.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 17066.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.451625,
    "mean_cost_per_repair_loop": 0.15054166666666666,
    "p95_cost_per_repair_loop": 0.18023,
    "cpsc": 0.15054166666666666,
    "realized_cpsc": 0.15054166666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.451625,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.15054166666666666,
    "mean_spend_per_solved_task": 0.15054166666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.15054166666666668,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.229142568087635e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 76510.33333333333,
    "tokens_per_success": 76510.33333333333,
    "conditional_tokens_among_solved_loops": 76510.33333333333,
    "mean_tokens_per_solved_task": 76510.33333333333,
    "mean_input_tokens_per_repair_loop": 69398.66666666667,
    "mean_output_tokens_per_repair_loop": 7111.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 47616.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.0,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 9.0,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 1.038214,
    "mean_cost_per_repair_loop": 0.34607133333333334,
    "p95_cost_per_repair_loop": 0.370568,
    "cpsc": 0.34607133333333334,
    "realized_cpsc": 0.34607133333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.038214,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.34607133333333334,
    "mean_spend_per_solved_task": 0.34607133333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.34607133333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 25281.0,
    "tokens_per_success": 25281.0,
    "conditional_tokens_among_solved_loops": 25281.0,
    "mean_tokens_per_solved_task": 25281.0,
    "mean_input_tokens_per_repair_loop": 21826.0,
    "mean_output_tokens_per_repair_loop": 3455.0,
    "mean_cache_read_tokens_per_repair_loop": 10069.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5024040000000001,
    "mean_cost_per_repair_loop": 0.16746800000000003,
    "p95_cost_per_repair_loop": 0.19421,
    "cpsc": 0.16746800000000003,
    "realized_cpsc": 0.16746800000000003,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5024040000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.16746800000000003,
    "mean_spend_per_solved_task": 0.16746800000000003,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.16746800000000003,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 61709.0,
    "tokens_per_success": 61709.0,
    "conditional_tokens_among_solved_loops": 61709.0,
    "mean_tokens_per_solved_task": 61709.0,
    "mean_input_tokens_per_repair_loop": 56671.666666666664,
    "mean_output_tokens_per_repair_loop": 5037.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 39765.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.666666666666666,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 10.666666666666666,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.766603,
    "mean_cost_per_repair_loop": 0.25553433333333336,
    "p95_cost_per_repair_loop": 0.288103,
    "cpsc": 0.25553433333333336,
    "realized_cpsc": 0.25553433333333336,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.766603,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.25553433333333336,
    "mean_spend_per_solved_task": 0.25553433333333336,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2555343333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": 1.448237255300536e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 24745.0,
    "tokens_per_success": 24745.0,
    "conditional_tokens_among_solved_loops": 24745.0,
    "mean_tokens_per_solved_task": 24745.0,
    "mean_input_tokens_per_repair_loop": 22584.0,
    "mean_output_tokens_per_repair_loop": 2161.0,
    "mean_cache_read_tokens_per_repair_loop": 12800.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.36045,
    "mean_cost_per_repair_loop": 0.12014999999999999,
    "p95_cost_per_repair_loop": 0.162509,
    "cpsc": 0.12014999999999999,
    "realized_cpsc": 0.12014999999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.36045,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12014999999999999,
    "mean_spend_per_solved_task": 0.12014999999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12014999999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 48544.0,
    "tokens_per_success": 48544.0,
    "conditional_tokens_among_solved_loops": 48544.0,
    "mean_tokens_per_solved_task": 48544.0,
    "mean_input_tokens_per_repair_loop": 44455.666666666664,
    "mean_output_tokens_per_repair_loop": 4088.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 24064.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.333333333333334,
    "p95_turns_per_repair_loop": 13.0,
    "mean_agent_steps_per_repair_loop": 10.333333333333334,
    "p95_agent_steps_per_repair_loop": 13.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.709921,
    "mean_cost_per_repair_loop": 0.23664033333333334,
    "p95_cost_per_repair_loop": 0.330547,
    "cpsc": 0.23664033333333334,
    "realized_cpsc": 0.23664033333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.709921,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.23664033333333334,
    "mean_spend_per_solved_task": 0.23664033333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2366403333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": 1.5638684087738729e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 22883.0,
    "tokens_per_success": 22883.0,
    "conditional_tokens_among_solved_loops": 22883.0,
    "mean_tokens_per_solved_task": 22883.0,
    "mean_input_tokens_per_repair_loop": 20823.0,
    "mean_output_tokens_per_repair_loop": 2060.0,
    "mean_cache_read_tokens_per_repair_loop": 9557.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.36872099999999997,
    "mean_cost_per_repair_loop": 0.12290699999999999,
    "p95_cost_per_repair_loop": 0.155827,
    "cpsc": 0.12290699999999999,
    "realized_cpsc": 0.12290699999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.36872099999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12290699999999999,
    "mean_spend_per_solved_task": 0.12290699999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12290699999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 57716.666666666664,
    "tokens_per_success": 57716.666666666664,
    "conditional_tokens_among_solved_loops": 57716.666666666664,
    "mean_tokens_per_solved_task": 57716.666666666664,
    "mean_input_tokens_per_repair_loop": 51934.333333333336,
    "mean_output_tokens_per_repair_loop": 5782.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 34304.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.836321,
    "mean_cost_per_repair_loop": 0.27877366666666664,
    "p95_cost_per_repair_loop": 0.329638,
    "cpsc": 0.27877366666666664,
    "realized_cpsc": 0.27877366666666664,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.836321,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.27877366666666664,
    "mean_spend_per_solved_task": 0.27877366666666664,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.27877366666666664,
    "mean_cost_delta_vs_gateway_per_repair_loop": 3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": 1.3275082469831041e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17108.333333333332,
    "tokens_per_success": 17108.333333333332,
    "conditional_tokens_among_solved_loops": 17108.333333333332,
    "mean_tokens_per_solved_task": 17108.333333333332,
    "mean_input_tokens_per_repair_loop": 15161.0,
    "mean_output_tokens_per_repair_loop": 1947.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 7680.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.298995,
    "mean_cost_per_repair_loop": 0.099665,
    "p95_cost_per_repair_loop": 0.104444,
    "cpsc": 0.099665,
    "realized_cpsc": 0.099665,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.298995,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.099665,
    "mean_spend_per_solved_task": 0.099665,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.099665,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 439494.6666666667,
    "tokens_per_success": 439494.6666666667,
    "conditional_tokens_among_solved_loops": 439494.6666666667,
    "mean_tokens_per_solved_task": 439494.6666666667,
    "mean_input_tokens_per_repair_loop": 426399.6666666667,
    "mean_output_tokens_per_repair_loop": 13095.0,
    "mean_cache_read_tokens_per_repair_loop": 348842.6666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 22.333333333333332,
    "p95_turns_per_repair_loop": 55.0,
    "mean_agent_steps_per_repair_loop": 22.333333333333332,
    "p95_agent_steps_per_repair_loop": 55.0,
    "mean_verifier_submissions_per_repair_loop": 3.3333333333333335,
    "mean_verifier_submissions_to_success": 3.3333333333333335,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 2.865169,
    "mean_cost_per_repair_loop": 0.9550563333333333,
    "p95_cost_per_repair_loop": 2.579999,
    "cpsc": 0.9550563333333333,
    "realized_cpsc": 0.9550563333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 2.865169,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.9550563333333333,
    "mean_spend_per_solved_task": 0.9550563333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.9550563333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.4802973661668753e-16,
    "cost_delta_vs_gateway_ratio": -1.5499581694834148e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 16094.0,
    "tokens_per_success": 16094.0,
    "conditional_tokens_among_solved_loops": 16094.0,
    "mean_tokens_per_solved_task": 16094.0,
    "mean_input_tokens_per_repair_loop": 13850.333333333334,
    "mean_output_tokens_per_repair_loop": 2243.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 9205.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 3687.6666666666665,
    "mean_reasoning_tokens_per_repair_loop": 406.3333333333333,
    "mean_turns_per_repair_loop": 5.333333333333333,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.333333333333333,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.059847449999999996,
    "mean_cost_per_repair_loop": 0.01994915,
    "p95_cost_per_repair_loop": 0.02338105,
    "cpsc": 0.01994915,
    "realized_cpsc": 0.01994915,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.059847449999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01994915,
    "mean_spend_per_solved_task": 0.01994915,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.019287233333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0006619166666666648,
    "cost_delta_vs_gateway_ratio": 0.03431890179514246
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 11246.666666666666,
    "tokens_per_success": 11246.666666666666,
    "conditional_tokens_among_solved_loops": 11246.666666666666,
    "mean_tokens_per_solved_task": 11246.666666666666,
    "mean_input_tokens_per_repair_loop": 9626.0,
    "mean_output_tokens_per_repair_loop": 1620.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 6036.0,
    "mean_cache_write_tokens_per_repair_loop": 2773.0,
    "mean_reasoning_tokens_per_repair_loop": 385.0,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.04383255,
    "mean_cost_per_repair_loop": 0.01461085,
    "p95_cost_per_repair_loop": 0.01526805,
    "cpsc": 0.01461085,
    "realized_cpsc": 0.01461085,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.04383255,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01461085,
    "mean_spend_per_solved_task": 0.01461085,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.01461085,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 42514.0,
    "tokens_per_success": 42514.0,
    "conditional_tokens_among_solved_loops": 42514.0,
    "mean_tokens_per_solved_task": 42514.0,
    "mean_input_tokens_per_repair_loop": 38327.0,
    "mean_output_tokens_per_repair_loop": 4187.0,
    "mean_cache_read_tokens_per_repair_loop": 31977.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 6329.333333333333,
    "mean_reasoning_tokens_per_repair_loop": 265.6666666666667,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.1087543,
    "mean_cost_per_repair_loop": 0.03625143333333333,
    "p95_cost_per_repair_loop": 0.04103855,
    "cpsc": 0.03625143333333333,
    "realized_cpsc": 0.03625143333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1087543,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.03625143333333333,
    "mean_spend_per_solved_task": 0.03625143333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03474276666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0015086666666666675,
    "cost_delta_vs_gateway_ratio": 0.04342390694274014
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 16166.333333333334,
    "tokens_per_success": 16166.333333333334,
    "conditional_tokens_among_solved_loops": 16166.333333333334,
    "mean_tokens_per_solved_task": 16166.333333333334,
    "mean_input_tokens_per_repair_loop": 14186.0,
    "mean_output_tokens_per_repair_loop": 1980.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 9448.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 3593.0,
    "mean_reasoning_tokens_per_repair_loop": 361.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.05538825,
    "mean_cost_per_repair_loop": 0.01846275,
    "p95_cost_per_repair_loop": 0.020572200000000002,
    "cpsc": 0.01846275,
    "realized_cpsc": 0.01846275,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.05538825,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01846275,
    "mean_spend_per_solved_task": 0.01846275,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.01846275,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 24939.0,
    "tokens_per_success": 24939.0,
    "conditional_tokens_among_solved_loops": 24939.0,
    "mean_tokens_per_solved_task": 24939.0,
    "mean_input_tokens_per_repair_loop": 23418.666666666668,
    "mean_output_tokens_per_repair_loop": 1520.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 17697.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 4902.333333333333,
    "mean_reasoning_tokens_per_repair_loop": 220.66666666666666,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.05351595000000001,
    "mean_cost_per_repair_loop": 0.01783865,
    "p95_cost_per_repair_loop": 0.0208364,
    "cpsc": 0.01783865,
    "realized_cpsc": 0.01783865,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.05351595000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01783865,
    "mean_spend_per_solved_task": 0.01783865,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.01783865,
    "mean_cost_delta_vs_gateway_per_repair_loop": 2.3129646346357427e-18,
    "cost_delta_vs_gateway_ratio": 1.2966029574187188e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 27535.0,
    "tokens_per_success": 27535.0,
    "conditional_tokens_among_solved_loops": 27535.0,
    "mean_tokens_per_solved_task": 27535.0,
    "mean_input_tokens_per_repair_loop": 25242.0,
    "mean_output_tokens_per_repair_loop": 2293.0,
    "mean_cache_read_tokens_per_repair_loop": 18043.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 6188.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 336.6666666666667,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0729245,
    "mean_cost_per_repair_loop": 0.02430816666666667,
    "p95_cost_per_repair_loop": 0.02736165,
    "cpsc": 0.02430816666666667,
    "realized_cpsc": 0.02430816666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0729245,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02430816666666667,
    "mean_spend_per_solved_task": 0.02430816666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.02430816666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17692.333333333332,
    "tokens_per_success": 17692.333333333332,
    "conditional_tokens_among_solved_loops": 17692.333333333332,
    "mean_tokens_per_solved_task": 17692.333333333332,
    "mean_input_tokens_per_repair_loop": 15890.0,
    "mean_output_tokens_per_repair_loop": 1802.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 10140.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 4831.333333333333,
    "mean_reasoning_tokens_per_repair_loop": 373.6666666666667,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0563557,
    "mean_cost_per_repair_loop": 0.018785233333333335,
    "p95_cost_per_repair_loop": 0.0252618,
    "cpsc": 0.018785233333333335,
    "realized_cpsc": 0.018785233333333335,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0563557,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.018785233333333335,
    "mean_spend_per_solved_task": 0.018785233333333335,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.01878523333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 2.3129646346357427e-18,
    "cost_delta_vs_gateway_ratio": 1.2312674501261148e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10570.666666666666,
    "tokens_per_success": 10570.666666666666,
    "conditional_tokens_among_solved_loops": 10570.666666666666,
    "mean_tokens_per_solved_task": 10570.666666666666,
    "mean_input_tokens_per_repair_loop": 9068.0,
    "mean_output_tokens_per_repair_loop": 1502.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 5396.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 2833.3333333333335,
    "mean_reasoning_tokens_per_repair_loop": 197.0,
    "mean_turns_per_repair_loop": 5.333333333333333,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.333333333333333,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.041805999999999996,
    "mean_cost_per_repair_loop": 0.013935333333333333,
    "p95_cost_per_repair_loop": 0.0163204,
    "cpsc": 0.013935333333333333,
    "realized_cpsc": 0.013935333333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.041805999999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.013935333333333333,
    "mean_spend_per_solved_task": 0.013935333333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.013935333333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -2.3129646346357427e-18,
    "cost_delta_vs_gateway_ratio": -1.6597842185110338e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 22875.666666666668,
    "tokens_per_success": 22875.666666666668,
    "conditional_tokens_among_solved_loops": 22875.666666666668,
    "mean_tokens_per_solved_task": 22875.666666666668,
    "mean_input_tokens_per_repair_loop": 21048.666666666668,
    "mean_output_tokens_per_repair_loop": 1827.0,
    "mean_cache_read_tokens_per_repair_loop": 14954.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 5293.333333333333,
    "mean_reasoning_tokens_per_repair_loop": 243.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0596253,
    "mean_cost_per_repair_loop": 0.0198751,
    "p95_cost_per_repair_loop": 0.02191605,
    "cpsc": 0.0198751,
    "realized_cpsc": 0.0198751,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0596253,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0198751,
    "mean_spend_per_solved_task": 0.0198751,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.019875100000000003,
    "mean_cost_delta_vs_gateway_per_repair_loop": -2.3129646346357427e-18,
    "cost_delta_vs_gateway_ratio": -1.1637499356661062e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 2,
    "solve_rate": 0.6666666666666666,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 68020.33333333333,
    "tokens_per_success": 102030.5,
    "conditional_tokens_among_solved_loops": 39097.5,
    "mean_tokens_per_solved_task": 39097.5,
    "mean_input_tokens_per_repair_loop": 64592.333333333336,
    "mean_output_tokens_per_repair_loop": 3428.0,
    "mean_cache_read_tokens_per_repair_loop": 55452.0,
    "mean_cache_write_tokens_per_repair_loop": 9105.333333333334,
    "mean_reasoning_tokens_per_repair_loop": 553.3333333333334,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 3.0,
    "mean_verifier_submissions_to_success": 1.5,
    "stop_reasons": {
      "agent_exit_repeatedformaterror": 1,
      "passed": 2
    },
    "total_model_spend_usd": 0.1125896,
    "mean_cost_per_repair_loop": 0.03752986666666667,
    "p95_cost_per_repair_loop": 0.052779599999999996,
    "cpsc": 0.0562948,
    "realized_cpsc": 0.0562948,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.1125896,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.029905,
    "mean_spend_per_solved_task": 0.029905,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.030629450000000003,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.006900416666666663,
    "cost_delta_vs_gateway_ratio": 0.2252869923118653
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 14349.0,
    "tokens_per_success": 14349.0,
    "conditional_tokens_among_solved_loops": 14349.0,
    "mean_tokens_per_solved_task": 14349.0,
    "mean_input_tokens_per_repair_loop": 12214.333333333334,
    "mean_output_tokens_per_repair_loop": 2134.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 7769.0,
    "mean_cache_write_tokens_per_repair_loop": 3490.3333333333335,
    "mean_reasoning_tokens_per_repair_loop": 370.3333333333333,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.05670845,
    "mean_cost_per_repair_loop": 0.018902816666666666,
    "p95_cost_per_repair_loop": 0.022013349999999998,
    "cpsc": 0.018902816666666666,
    "realized_cpsc": 0.018902816666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.05670845,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.018902816666666666,
    "mean_spend_per_solved_task": 0.018902816666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.018902816666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23459.0,
    "tokens_per_success": 23459.0,
    "conditional_tokens_among_solved_loops": 23459.0,
    "mean_tokens_per_solved_task": 23459.0,
    "mean_input_tokens_per_repair_loop": 20982.333333333332,
    "mean_output_tokens_per_repair_loop": 2476.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 14440.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 5610.0,
    "mean_reasoning_tokens_per_repair_loop": 332.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0727456,
    "mean_cost_per_repair_loop": 0.024248533333333332,
    "p95_cost_per_repair_loop": 0.025593599999999998,
    "cpsc": 0.024248533333333332,
    "realized_cpsc": 0.024248533333333332,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0727456,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.024248533333333332,
    "mean_spend_per_solved_task": 0.024248533333333332,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.024248533333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 8714.666666666666,
    "tokens_per_success": 8714.666666666666,
    "conditional_tokens_among_solved_loops": 8714.666666666666,
    "mean_tokens_per_solved_task": 8714.666666666666,
    "mean_input_tokens_per_repair_loop": 7534.0,
    "mean_output_tokens_per_repair_loop": 1180.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 3938.0,
    "mean_cache_write_tokens_per_repair_loop": 2574.0,
    "mean_reasoning_tokens_per_repair_loop": 132.0,
    "mean_turns_per_repair_loop": 4.333333333333333,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.333333333333333,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0351519,
    "mean_cost_per_repair_loop": 0.0117173,
    "p95_cost_per_repair_loop": 0.01280975,
    "cpsc": 0.0117173,
    "realized_cpsc": 0.0117173,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0351519,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0117173,
    "mean_spend_per_solved_task": 0.0117173,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0117173,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17002.0,
    "tokens_per_success": 17002.0,
    "conditional_tokens_among_solved_loops": 17002.0,
    "mean_tokens_per_solved_task": 17002.0,
    "mean_input_tokens_per_repair_loop": 15055.333333333334,
    "mean_output_tokens_per_repair_loop": 1946.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 10033.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 3842.3333333333335,
    "mean_reasoning_tokens_per_repair_loop": 335.3333333333333,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.05599775,
    "mean_cost_per_repair_loop": 0.018665916666666667,
    "p95_cost_per_repair_loop": 0.02022785,
    "cpsc": 0.018665916666666667,
    "realized_cpsc": 0.018665916666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.05599775,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.018665916666666667,
    "mean_spend_per_solved_task": 0.018665916666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.018665916666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 13988.333333333334,
    "tokens_per_success": 13988.333333333334,
    "conditional_tokens_among_solved_loops": 13988.333333333334,
    "mean_tokens_per_solved_task": 13988.333333333334,
    "mean_input_tokens_per_repair_loop": 12518.333333333334,
    "mean_output_tokens_per_repair_loop": 1470.0,
    "mean_cache_read_tokens_per_repair_loop": 8554.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 3147.0,
    "mean_reasoning_tokens_per_repair_loop": 232.33333333333334,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.04327855,
    "mean_cost_per_repair_loop": 0.014426183333333334,
    "p95_cost_per_repair_loop": 0.01696355,
    "cpsc": 0.014426183333333334,
    "realized_cpsc": 0.014426183333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.04327855,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.014426183333333334,
    "mean_spend_per_solved_task": 0.014426183333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.014426183333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 43670.0,
    "tokens_per_success": 43670.0,
    "conditional_tokens_among_solved_loops": 43670.0,
    "mean_tokens_per_solved_task": 43670.0,
    "mean_input_tokens_per_repair_loop": 40405.333333333336,
    "mean_output_tokens_per_repair_loop": 3264.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 33627.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 6757.0,
    "mean_reasoning_tokens_per_repair_loop": 238.33333333333334,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.09425395,
    "mean_cost_per_repair_loop": 0.03141798333333334,
    "p95_cost_per_repair_loop": 0.0321461,
    "cpsc": 0.03141798333333334,
    "realized_cpsc": 0.03141798333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.09425395,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.03141798333333334,
    "mean_spend_per_solved_task": 0.03141798333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03141798333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 4.625929269271485e-18,
    "cost_delta_vs_gateway_ratio": 1.4723826224592666e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 8333.333333333334,
    "tokens_per_success": 8333.333333333334,
    "conditional_tokens_among_solved_loops": 8333.333333333334,
    "mean_tokens_per_solved_task": 8333.333333333334,
    "mean_input_tokens_per_repair_loop": 7054.333333333333,
    "mean_output_tokens_per_repair_loop": 1279.0,
    "mean_cache_read_tokens_per_repair_loop": 3515.6666666666665,
    "mean_cache_write_tokens_per_repair_loop": 2509.6666666666665,
    "mean_reasoning_tokens_per_repair_loop": 148.0,
    "mean_turns_per_repair_loop": 4.0,
    "p95_turns_per_repair_loop": 4.0,
    "mean_agent_steps_per_repair_loop": 4.0,
    "p95_agent_steps_per_repair_loop": 4.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.03657495,
    "mean_cost_per_repair_loop": 0.01219165,
    "p95_cost_per_repair_loop": 0.012928,
    "cpsc": 0.01219165,
    "realized_cpsc": 0.01219165,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.03657495,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01219165,
    "mean_spend_per_solved_task": 0.01219165,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.01219165,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-luna[low]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15484.0,
    "tokens_per_success": 15484.0,
    "conditional_tokens_among_solved_loops": 15484.0,
    "mean_tokens_per_solved_task": 15484.0,
    "mean_input_tokens_per_repair_loop": 13564.666666666666,
    "mean_output_tokens_per_repair_loop": 1919.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 9324.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 3275.3333333333335,
    "mean_reasoning_tokens_per_repair_loop": 385.3333333333333,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0525228,
    "mean_cost_per_repair_loop": 0.0175076,
    "p95_cost_per_repair_loop": 0.01857195,
    "cpsc": 0.0175076,
    "realized_cpsc": 0.0175076,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0525228,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0175076,
    "mean_spend_per_solved_task": 0.0175076,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0175076,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 21193.666666666668,
    "tokens_per_success": 21193.666666666668,
    "conditional_tokens_among_solved_loops": 21193.666666666668,
    "mean_tokens_per_solved_task": 21193.666666666668,
    "mean_input_tokens_per_repair_loop": 18040.666666666668,
    "mean_output_tokens_per_repair_loop": 3153.0,
    "mean_cache_read_tokens_per_repair_loop": 12009.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 5072.0,
    "mean_reasoning_tokens_per_repair_loop": 410.6666666666667,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.4112695,
    "mean_cost_per_repair_loop": 0.13708983333333333,
    "p95_cost_per_repair_loop": 0.143303,
    "cpsc": 0.13708983333333333,
    "realized_cpsc": 0.13708983333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.4112695,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.13708983333333333,
    "mean_spend_per_solved_task": 0.13708983333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.13708983333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": 1.349751227145651e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23495.333333333332,
    "tokens_per_success": 23495.333333333332,
    "conditional_tokens_among_solved_loops": 23495.333333333332,
    "mean_tokens_per_solved_task": 23495.333333333332,
    "mean_input_tokens_per_repair_loop": 20326.0,
    "mean_output_tokens_per_repair_loop": 3169.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 13781.0,
    "mean_cache_write_tokens_per_repair_loop": 5722.0,
    "mean_reasoning_tokens_per_repair_loop": 803.0,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.42554400000000003,
    "mean_cost_per_repair_loop": 0.141848,
    "p95_cost_per_repair_loop": 0.16818925,
    "cpsc": 0.141848,
    "realized_cpsc": 0.141848,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.42554400000000003,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.141848,
    "mean_spend_per_solved_task": 0.141848,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.141848,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 53439.666666666664,
    "tokens_per_success": 53439.666666666664,
    "conditional_tokens_among_solved_loops": 53439.666666666664,
    "mean_tokens_per_solved_task": 53439.666666666664,
    "mean_input_tokens_per_repair_loop": 47547.0,
    "mean_output_tokens_per_repair_loop": 5892.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 38170.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 9356.666666666666,
    "mean_reasoning_tokens_per_repair_loop": 444.0,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.7633329999999999,
    "mean_cost_per_repair_loop": 0.25444433333333333,
    "p95_cost_per_repair_loop": 0.2702785,
    "cpsc": 0.25444433333333333,
    "realized_cpsc": 0.25444433333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.7633329999999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.25444433333333333,
    "mean_spend_per_solved_task": 0.25444433333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.25444433333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": -3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": -1.4544412787409381e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 22494.333333333332,
    "tokens_per_success": 22494.333333333332,
    "conditional_tokens_among_solved_loops": 22494.333333333332,
    "mean_tokens_per_solved_task": 22494.333333333332,
    "mean_input_tokens_per_repair_loop": 19517.333333333332,
    "mean_output_tokens_per_repair_loop": 2977.0,
    "mean_cache_read_tokens_per_repair_loop": 13333.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 5355.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 667.0,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.40076924999999997,
    "mean_cost_per_repair_loop": 0.13358974999999998,
    "p95_cost_per_repair_loop": 0.138819,
    "cpsc": 0.13358974999999998,
    "realized_cpsc": 0.13358974999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.40076924999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.13358974999999998,
    "mean_spend_per_solved_task": 0.13358974999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.13358974999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 28281.333333333332,
    "tokens_per_success": 28281.333333333332,
    "conditional_tokens_among_solved_loops": 28281.333333333332,
    "mean_tokens_per_solved_task": 28281.333333333332,
    "mean_input_tokens_per_repair_loop": 25927.0,
    "mean_output_tokens_per_repair_loop": 2354.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 18712.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 6396.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 322.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.372166,
    "mean_cost_per_repair_loop": 0.12405533333333334,
    "p95_cost_per_repair_loop": 0.13076075,
    "cpsc": 0.12405533333333334,
    "realized_cpsc": 0.12405533333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.372166,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12405533333333334,
    "mean_spend_per_solved_task": 0.12405533333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12405533333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 49719.0,
    "tokens_per_success": 49719.0,
    "conditional_tokens_among_solved_loops": 49719.0,
    "mean_tokens_per_solved_task": 49719.0,
    "mean_input_tokens_per_repair_loop": 45845.666666666664,
    "mean_output_tokens_per_repair_loop": 3873.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 33894.0,
    "mean_cache_write_tokens_per_repair_loop": 10939.666666666666,
    "mean_reasoning_tokens_per_repair_loop": 447.3333333333333,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.6197397499999999,
    "mean_cost_per_repair_loop": 0.20657991666666664,
    "p95_cost_per_repair_loop": 0.25332825,
    "cpsc": 0.20657991666666664,
    "realized_cpsc": 0.20657991666666664,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.6197397499999999,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.20657991666666664,
    "mean_spend_per_solved_task": 0.20657991666666664,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.20657991666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -3.700743415417188e-17,
    "cost_delta_vs_gateway_ratio": -1.7914342667630348e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 21087.333333333332,
    "tokens_per_success": 21087.333333333332,
    "conditional_tokens_among_solved_loops": 21087.333333333332,
    "mean_tokens_per_solved_task": 21087.333333333332,
    "mean_input_tokens_per_repair_loop": 18767.666666666668,
    "mean_output_tokens_per_repair_loop": 2319.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 11956.0,
    "mean_cache_write_tokens_per_repair_loop": 5892.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 444.3333333333333,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.35097649999999997,
    "mean_cost_per_repair_loop": 0.11699216666666666,
    "p95_cost_per_repair_loop": 0.14209175,
    "cpsc": 0.11699216666666666,
    "realized_cpsc": 0.11699216666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.35097649999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11699216666666666,
    "mean_spend_per_solved_task": 0.11699216666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11699216666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.581620172041656e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 19132.0,
    "tokens_per_success": 19132.0,
    "conditional_tokens_among_solved_loops": 19132.0,
    "mean_tokens_per_solved_task": 19132.0,
    "mean_input_tokens_per_repair_loop": 16881.666666666668,
    "mean_output_tokens_per_repair_loop": 2250.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 10740.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 5300.0,
    "mean_reasoning_tokens_per_repair_loop": 236.0,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.330631,
    "mean_cost_per_repair_loop": 0.11021033333333334,
    "p95_cost_per_repair_loop": 0.12026675,
    "cpsc": 0.11021033333333334,
    "realized_cpsc": 0.11021033333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.330631,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11021033333333334,
    "mean_spend_per_solved_task": 0.11021033333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11021033333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 37231.0,
    "tokens_per_success": 37231.0,
    "conditional_tokens_among_solved_loops": 37231.0,
    "mean_tokens_per_solved_task": 37231.0,
    "mean_input_tokens_per_repair_loop": 34479.666666666664,
    "mean_output_tokens_per_repair_loop": 2751.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 25547.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 7809.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 389.0,
    "mean_turns_per_repair_loop": 8.0,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.0,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.44920775,
    "mean_cost_per_repair_loop": 0.14973591666666666,
    "p95_cost_per_repair_loop": 0.1544935,
    "cpsc": 0.14973591666666666,
    "realized_cpsc": 0.14973591666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.44920775,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14973591666666666,
    "mean_spend_per_solved_task": 0.14973591666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14973591666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 37447.333333333336,
    "tokens_per_success": 37447.333333333336,
    "conditional_tokens_among_solved_loops": 37447.333333333336,
    "mean_tokens_per_solved_task": 37447.333333333336,
    "mean_input_tokens_per_repair_loop": 33149.666666666664,
    "mean_output_tokens_per_repair_loop": 4297.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 25467.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 7663.333333333333,
    "mean_reasoning_tokens_per_repair_loop": 470.3333333333333,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5689635,
    "mean_cost_per_repair_loop": 0.18965449999999998,
    "p95_cost_per_repair_loop": 0.2132245,
    "cpsc": 0.18965449999999998,
    "realized_cpsc": 0.18965449999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5689635,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.18965449999999998,
    "mean_spend_per_solved_task": 0.18965449999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.18965449999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 22635.0,
    "tokens_per_success": 22635.0,
    "conditional_tokens_among_solved_loops": 22635.0,
    "mean_tokens_per_solved_task": 22635.0,
    "mean_input_tokens_per_repair_loop": 19320.0,
    "mean_output_tokens_per_repair_loop": 3315.0,
    "mean_cache_read_tokens_per_repair_loop": 13255.0,
    "mean_cache_write_tokens_per_repair_loop": 5107.0,
    "mean_reasoning_tokens_per_repair_loop": 717.3333333333334,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.42835875,
    "mean_cost_per_repair_loop": 0.14278625,
    "p95_cost_per_repair_loop": 0.15890025,
    "cpsc": 0.14278625,
    "realized_cpsc": 0.14278625,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.42835875,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14278625,
    "mean_spend_per_solved_task": 0.14278625,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14278625,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 30359.0,
    "tokens_per_success": 30359.0,
    "conditional_tokens_among_solved_loops": 30359.0,
    "mean_tokens_per_solved_task": 30359.0,
    "mean_input_tokens_per_repair_loop": 27283.666666666668,
    "mean_output_tokens_per_repair_loop": 3075.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 19556.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 6794.0,
    "mean_reasoning_tokens_per_repair_loop": 502.6666666666667,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.4474975,
    "mean_cost_per_repair_loop": 0.14916583333333333,
    "p95_cost_per_repair_loop": 0.1595685,
    "cpsc": 0.14916583333333333,
    "realized_cpsc": 0.14916583333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.4474975,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.14916583333333333,
    "mean_spend_per_solved_task": 0.14916583333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.14916583333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15075.333333333334,
    "tokens_per_success": 15075.333333333334,
    "conditional_tokens_among_solved_loops": 15075.333333333334,
    "mean_tokens_per_solved_task": 15075.333333333334,
    "mean_input_tokens_per_repair_loop": 13076.333333333334,
    "mean_output_tokens_per_repair_loop": 1999.0,
    "mean_cache_read_tokens_per_repair_loop": 8395.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 3653.6666666666665,
    "mean_reasoning_tokens_per_repair_loop": 185.66666666666666,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.27641475,
    "mean_cost_per_repair_loop": 0.09213824999999999,
    "p95_cost_per_repair_loop": 0.0949025,
    "cpsc": 0.09213824999999999,
    "realized_cpsc": 0.09213824999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.27641475,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09213824999999999,
    "mean_spend_per_solved_task": 0.09213824999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.09213824999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 29624.666666666668,
    "tokens_per_success": 29624.666666666668,
    "conditional_tokens_among_solved_loops": 29624.666666666668,
    "mean_tokens_per_solved_task": 29624.666666666668,
    "mean_input_tokens_per_repair_loop": 26007.0,
    "mean_output_tokens_per_repair_loop": 3617.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 17446.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 6709.333333333333,
    "mean_reasoning_tokens_per_repair_loop": 480.6666666666667,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.5053295,
    "mean_cost_per_repair_loop": 0.16844316666666667,
    "p95_cost_per_repair_loop": 0.17230075,
    "cpsc": 0.16844316666666667,
    "realized_cpsc": 0.16844316666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.5053295,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.16844316666666667,
    "mean_spend_per_solved_task": 0.16844316666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.16844316666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 19426.666666666668,
    "tokens_per_success": 19426.666666666668,
    "conditional_tokens_among_solved_loops": 19426.666666666668,
    "mean_tokens_per_solved_task": 19426.666666666668,
    "mean_input_tokens_per_repair_loop": 16875.0,
    "mean_output_tokens_per_repair_loop": 2551.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 11453.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 4602.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 420.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.345415,
    "mean_cost_per_repair_loop": 0.11513833333333334,
    "p95_cost_per_repair_loop": 0.119962,
    "cpsc": 0.11513833333333334,
    "realized_cpsc": 0.11513833333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.345415,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11513833333333334,
    "mean_spend_per_solved_task": 0.11513833333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11513833333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 46196.333333333336,
    "tokens_per_success": 46196.333333333336,
    "conditional_tokens_among_solved_loops": 46196.333333333336,
    "mean_tokens_per_solved_task": 46196.333333333336,
    "mean_input_tokens_per_repair_loop": 40661.0,
    "mean_output_tokens_per_repair_loop": 5535.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 31091.0,
    "mean_cache_write_tokens_per_repair_loop": 9552.0,
    "mean_reasoning_tokens_per_repair_loop": 563.0,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.7241865,
    "mean_cost_per_repair_loop": 0.24139549999999999,
    "p95_cost_per_repair_loop": 0.262808,
    "cpsc": 0.24139549999999999,
    "realized_cpsc": 0.24139549999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.7241865,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.24139549999999999,
    "mean_spend_per_solved_task": 0.24139549999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.24139549999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17028.0,
    "tokens_per_success": 17028.0,
    "conditional_tokens_among_solved_loops": 17028.0,
    "mean_tokens_per_solved_task": 17028.0,
    "mean_input_tokens_per_repair_loop": 15184.333333333334,
    "mean_output_tokens_per_repair_loop": 1843.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 9408.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 4742.0,
    "mean_reasoning_tokens_per_repair_loop": 160.66666666666666,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.28446499999999997,
    "mean_cost_per_repair_loop": 0.09482166666666665,
    "p95_cost_per_repair_loop": 0.1128065,
    "cpsc": 0.09482166666666665,
    "realized_cpsc": 0.09482166666666665,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.28446499999999997,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09482166666666665,
    "mean_spend_per_solved_task": 0.09482166666666665,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.09482166666666665,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 21669.666666666668,
    "tokens_per_success": 21669.666666666668,
    "conditional_tokens_among_solved_loops": 21669.666666666668,
    "mean_tokens_per_solved_task": 21669.666666666668,
    "mean_input_tokens_per_repair_loop": 18366.666666666668,
    "mean_output_tokens_per_repair_loop": 3303.0,
    "mean_cache_read_tokens_per_repair_loop": 12254.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 5147.333333333333,
    "mean_reasoning_tokens_per_repair_loop": 607.3333333333334,
    "mean_turns_per_repair_loop": 6.333333333333333,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.333333333333333,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.426639,
    "mean_cost_per_repair_loop": 0.142213,
    "p95_cost_per_repair_loop": 0.147216,
    "cpsc": 0.142213,
    "realized_cpsc": 0.142213,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.426639,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.142213,
    "mean_spend_per_solved_task": 0.142213,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.142213,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17947.666666666668,
    "tokens_per_success": 17947.666666666668,
    "conditional_tokens_among_solved_loops": 17947.666666666668,
    "mean_tokens_per_solved_task": 17947.666666666668,
    "mean_input_tokens_per_repair_loop": 14709.666666666666,
    "mean_output_tokens_per_repair_loop": 3238.0,
    "mean_cache_read_tokens_per_repair_loop": 8983.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 4769.333333333333,
    "mean_reasoning_tokens_per_repair_loop": 820.6666666666666,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.2043375,
    "mean_cost_per_repair_loop": 0.0681125,
    "p95_cost_per_repair_loop": 0.070218375,
    "cpsc": 0.0681125,
    "realized_cpsc": 0.0681125,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.2043375,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0681125,
    "mean_spend_per_solved_task": 0.0681125,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0681125,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23561.0,
    "tokens_per_success": 23561.0,
    "conditional_tokens_among_solved_loops": 23561.0,
    "mean_tokens_per_solved_task": 23561.0,
    "mean_input_tokens_per_repair_loop": 20484.0,
    "mean_output_tokens_per_repair_loop": 3077.0,
    "mean_cache_read_tokens_per_repair_loop": 14299.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 5024.0,
    "mean_reasoning_tokens_per_repair_loop": 868.0,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.20499450000000002,
    "mean_cost_per_repair_loop": 0.0683315,
    "p95_cost_per_repair_loop": 0.077978375,
    "cpsc": 0.0683315,
    "realized_cpsc": 0.0683315,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.20499450000000002,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0683315,
    "mean_spend_per_solved_task": 0.0683315,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0683315,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 56041.333333333336,
    "tokens_per_success": 56041.333333333336,
    "conditional_tokens_among_solved_loops": 56041.333333333336,
    "mean_tokens_per_solved_task": 56041.333333333336,
    "mean_input_tokens_per_repair_loop": 49447.0,
    "mean_output_tokens_per_repair_loop": 6594.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 39793.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 9633.666666666666,
    "mean_reasoning_tokens_per_repair_loop": 803.3333333333334,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.417055625,
    "mean_cost_per_repair_loop": 0.13901854166666666,
    "p95_cost_per_repair_loop": 0.168872875,
    "cpsc": 0.13901854166666666,
    "realized_cpsc": 0.13901854166666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.417055625,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.13901854166666666,
    "mean_spend_per_solved_task": 0.13901854166666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.13901854166666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 25828.0,
    "tokens_per_success": 25828.0,
    "conditional_tokens_among_solved_loops": 25828.0,
    "mean_tokens_per_solved_task": 25828.0,
    "mean_input_tokens_per_repair_loop": 22600.666666666668,
    "mean_output_tokens_per_repair_loop": 3227.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 15699.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 5116.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 901.0,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.218356,
    "mean_cost_per_repair_loop": 0.07278533333333333,
    "p95_cost_per_repair_loop": 0.08043325,
    "cpsc": 0.07278533333333333,
    "realized_cpsc": 0.07278533333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.218356,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07278533333333333,
    "mean_spend_per_solved_task": 0.07278533333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07278533333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 37537.333333333336,
    "tokens_per_success": 37537.333333333336,
    "conditional_tokens_among_solved_loops": 37537.333333333336,
    "mean_tokens_per_solved_task": 37537.333333333336,
    "mean_input_tokens_per_repair_loop": 34637.0,
    "mean_output_tokens_per_repair_loop": 2900.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 26173.0,
    "mean_cache_write_tokens_per_repair_loop": 7644.0,
    "mean_reasoning_tokens_per_repair_loop": 624.6666666666666,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.22795725,
    "mean_cost_per_repair_loop": 0.07598575,
    "p95_cost_per_repair_loop": 0.08286475,
    "cpsc": 0.07598575,
    "realized_cpsc": 0.07598575,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.22795725,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07598575,
    "mean_spend_per_solved_task": 0.07598575,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07598575,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 40777.0,
    "tokens_per_success": 40777.0,
    "conditional_tokens_among_solved_loops": 40777.0,
    "mean_tokens_per_solved_task": 40777.0,
    "mean_input_tokens_per_repair_loop": 36691.666666666664,
    "mean_output_tokens_per_repair_loop": 4085.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 25303.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 10377.0,
    "mean_reasoning_tokens_per_repair_loop": 800.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.307684625,
    "mean_cost_per_repair_loop": 0.10256154166666666,
    "p95_cost_per_repair_loop": 0.11120775,
    "cpsc": 0.10256154166666666,
    "realized_cpsc": 0.10256154166666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.307684625,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.10256154166666666,
    "mean_spend_per_solved_task": 0.10256154166666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.10256154166666669,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.8041574625725228e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15467.333333333334,
    "tokens_per_success": 15467.333333333334,
    "conditional_tokens_among_solved_loops": 15467.333333333334,
    "mean_tokens_per_solved_task": 15467.333333333334,
    "mean_input_tokens_per_repair_loop": 13049.333333333334,
    "mean_output_tokens_per_repair_loop": 2418.0,
    "mean_cache_read_tokens_per_repair_loop": 8285.0,
    "mean_cache_write_tokens_per_repair_loop": 3846.3333333333335,
    "mean_reasoning_tokens_per_repair_loop": 663.3333333333334,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.15796812500000001,
    "mean_cost_per_repair_loop": 0.052656041666666674,
    "p95_cost_per_repair_loop": 0.055872375,
    "cpsc": 0.052656041666666674,
    "realized_cpsc": 0.052656041666666674,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.15796812500000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.052656041666666674,
    "mean_spend_per_solved_task": 0.052656041666666674,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.05265604166666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 9.25185853854297e-18,
    "cost_delta_vs_gateway_ratio": 1.7570364664155452e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 16272.0,
    "tokens_per_success": 16272.0,
    "conditional_tokens_among_solved_loops": 16272.0,
    "mean_tokens_per_solved_task": 16272.0,
    "mean_input_tokens_per_repair_loop": 14128.0,
    "mean_output_tokens_per_repair_loop": 2144.0,
    "mean_cache_read_tokens_per_repair_loop": 9525.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 3760.3333333333335,
    "mean_reasoning_tokens_per_repair_loop": 322.3333333333333,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.145192375,
    "mean_cost_per_repair_loop": 0.04839745833333334,
    "p95_cost_per_repair_loop": 0.052380375,
    "cpsc": 0.04839745833333334,
    "realized_cpsc": 0.04839745833333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.145192375,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.04839745833333334,
    "mean_spend_per_solved_task": 0.04839745833333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04839745833333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 9.25185853854297e-18,
    "cost_delta_vs_gateway_ratio": 1.91164140786518e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 25933.333333333332,
    "tokens_per_success": 25933.333333333332,
    "conditional_tokens_among_solved_loops": 25933.333333333332,
    "mean_tokens_per_solved_task": 25933.333333333332,
    "mean_input_tokens_per_repair_loop": 23549.0,
    "mean_output_tokens_per_repair_loop": 2384.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 15769.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 6036.0,
    "mean_reasoning_tokens_per_repair_loop": 478.3333333333333,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.188787,
    "mean_cost_per_repair_loop": 0.062929,
    "p95_cost_per_repair_loop": 0.06663275,
    "cpsc": 0.062929,
    "realized_cpsc": 0.062929,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.188787,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.062929,
    "mean_spend_per_solved_task": 0.062929,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.062929,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 40629.666666666664,
    "tokens_per_success": 40629.666666666664,
    "conditional_tokens_among_solved_loops": 40629.666666666664,
    "mean_tokens_per_solved_task": 40629.666666666664,
    "mean_input_tokens_per_repair_loop": 35941.333333333336,
    "mean_output_tokens_per_repair_loop": 4688.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 27444.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 8478.0,
    "mean_reasoning_tokens_per_repair_loop": 888.3333333333334,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.31118199999999996,
    "mean_cost_per_repair_loop": 0.10372733333333332,
    "p95_cost_per_repair_loop": 0.124178875,
    "cpsc": 0.10372733333333332,
    "realized_cpsc": 0.10372733333333332,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.31118199999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.10372733333333332,
    "mean_spend_per_solved_task": 0.10372733333333332,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.09879420833333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.004933124999999983,
    "cost_delta_vs_gateway_ratio": 0.04993334207766041
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 29576.666666666668,
    "tokens_per_success": 29576.666666666668,
    "conditional_tokens_among_solved_loops": 29576.666666666668,
    "mean_tokens_per_solved_task": 29576.666666666668,
    "mean_input_tokens_per_repair_loop": 25837.666666666668,
    "mean_output_tokens_per_repair_loop": 3739.0,
    "mean_cache_read_tokens_per_repair_loop": 18218.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 5611.0,
    "mean_reasoning_tokens_per_repair_loop": 1059.3333333333333,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.3333333333333333,
    "mean_verifier_submissions_to_success": 1.3333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.249582125,
    "mean_cost_per_repair_loop": 0.08319404166666666,
    "p95_cost_per_repair_loop": 0.1135535,
    "cpsc": 0.08319404166666666,
    "realized_cpsc": 0.08319404166666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.249582125,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.08319404166666666,
    "mean_spend_per_solved_task": 0.08319404166666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.08319404166666668,
    "mean_cost_delta_vs_gateway_per_repair_loop": -9.25185853854297e-18,
    "cost_delta_vs_gateway_ratio": -1.1120818694699755e-16
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 2,
    "solve_rate": 0.6666666666666666,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 104545.0,
    "tokens_per_success": 156817.5,
    "conditional_tokens_among_solved_loops": 45894.0,
    "mean_tokens_per_solved_task": 45894.0,
    "mean_input_tokens_per_repair_loop": 98566.33333333333,
    "mean_output_tokens_per_repair_loop": 5978.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 87588.0,
    "mean_cache_write_tokens_per_repair_loop": 10024.333333333334,
    "mean_reasoning_tokens_per_repair_loop": 2091.3333333333335,
    "mean_turns_per_repair_loop": 11.666666666666666,
    "p95_turns_per_repair_loop": 17.0,
    "mean_agent_steps_per_repair_loop": 11.666666666666666,
    "p95_agent_steps_per_repair_loop": 17.0,
    "mean_verifier_submissions_per_repair_loop": 3.0,
    "mean_verifier_submissions_to_success": 1.5,
    "stop_reasons": {
      "agent_exit_repeatedformaterror": 1,
      "passed": 2
    },
    "total_model_spend_usd": 0.435864125,
    "mean_cost_per_repair_loop": 0.14528804166666667,
    "p95_cost_per_repair_loop": 0.230238375,
    "cpsc": 0.2179320625,
    "realized_cpsc": 0.2179320625,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.435864125,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.102812875,
    "mean_spend_per_solved_task": 0.102812875,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.12156220833333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.023725833333333318,
    "cost_delta_vs_gateway_ratio": 0.1951744185847231
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 13667.666666666666,
    "tokens_per_success": 13667.666666666666,
    "conditional_tokens_among_solved_loops": 13667.666666666666,
    "mean_tokens_per_solved_task": 13667.666666666666,
    "mean_input_tokens_per_repair_loop": 11719.333333333334,
    "mean_output_tokens_per_repair_loop": 1948.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 7341.666666666667,
    "mean_cache_write_tokens_per_repair_loop": 3351.6666666666665,
    "mean_reasoning_tokens_per_repair_loop": 231.33333333333334,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.132298125,
    "mean_cost_per_repair_loop": 0.044099374999999996,
    "p95_cost_per_repair_loop": 0.050202375,
    "cpsc": 0.044099374999999996,
    "realized_cpsc": 0.044099374999999996,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.132298125,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.044099374999999996,
    "mean_spend_per_solved_task": 0.044099374999999996,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.044099374999999996,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 30002.0,
    "tokens_per_success": 30002.0,
    "conditional_tokens_among_solved_loops": 30002.0,
    "mean_tokens_per_solved_task": 30002.0,
    "mean_input_tokens_per_repair_loop": 26748.0,
    "mean_output_tokens_per_repair_loop": 3254.0,
    "mean_cache_read_tokens_per_repair_loop": 18372.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 6544.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 590.6666666666666,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.23529575,
    "mean_cost_per_repair_loop": 0.07843191666666667,
    "p95_cost_per_repair_loop": 0.094600875,
    "cpsc": 0.07843191666666667,
    "realized_cpsc": 0.07843191666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.23529575,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07843191666666667,
    "mean_spend_per_solved_task": 0.07843191666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07843191666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 19768.0,
    "tokens_per_success": 19768.0,
    "conditional_tokens_among_solved_loops": 19768.0,
    "mean_tokens_per_solved_task": 19768.0,
    "mean_input_tokens_per_repair_loop": 16886.0,
    "mean_output_tokens_per_repair_loop": 2882.0,
    "mean_cache_read_tokens_per_repair_loop": 10811.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 4312.666666666667,
    "mean_reasoning_tokens_per_repair_loop": 836.3333333333334,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.19144475,
    "mean_cost_per_repair_loop": 0.06381491666666667,
    "p95_cost_per_repair_loop": 0.073673625,
    "cpsc": 0.06381491666666667,
    "realized_cpsc": 0.06381491666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.19144475,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.06381491666666667,
    "mean_spend_per_solved_task": 0.06381491666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06381491666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 54974.666666666664,
    "tokens_per_success": 54974.666666666664,
    "conditional_tokens_among_solved_loops": 54974.666666666664,
    "mean_tokens_per_solved_task": 54974.666666666664,
    "mean_input_tokens_per_repair_loop": 48620.666666666664,
    "mean_output_tokens_per_repair_loop": 6354.0,
    "mean_cache_read_tokens_per_repair_loop": 38424.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 10176.0,
    "mean_reasoning_tokens_per_repair_loop": 1006.3333333333334,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.4102985,
    "mean_cost_per_repair_loop": 0.13676616666666666,
    "p95_cost_per_repair_loop": 0.147829125,
    "cpsc": 0.13676616666666666,
    "realized_cpsc": 0.13676616666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.4102985,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.13676616666666666,
    "mean_spend_per_solved_task": 0.13676616666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.13676616666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 14717.0,
    "tokens_per_success": 14717.0,
    "conditional_tokens_among_solved_loops": 14717.0,
    "mean_tokens_per_solved_task": 14717.0,
    "mean_input_tokens_per_repair_loop": 12904.666666666666,
    "mean_output_tokens_per_repair_loop": 1812.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 8432.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 3437.3333333333335,
    "mean_reasoning_tokens_per_repair_loop": 204.66666666666666,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.12786675,
    "mean_cost_per_repair_loop": 0.04262225,
    "p95_cost_per_repair_loop": 0.045208,
    "cpsc": 0.04262225,
    "realized_cpsc": 0.04262225,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.12786675,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.04262225,
    "mean_spend_per_solved_task": 0.04262225,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04262225,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-terra[low]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 18369.666666666668,
    "tokens_per_success": 18369.666666666668,
    "conditional_tokens_among_solved_loops": 18369.666666666668,
    "mean_tokens_per_solved_task": 18369.666666666668,
    "mean_input_tokens_per_repair_loop": 15371.333333333334,
    "mean_output_tokens_per_repair_loop": 2998.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 9574.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 4833.0,
    "mean_reasoning_tokens_per_repair_loop": 804.3333333333334,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.194645125,
    "mean_cost_per_repair_loop": 0.06488170833333333,
    "p95_cost_per_repair_loop": 0.067970375,
    "cpsc": 0.06488170833333333,
    "realized_cpsc": 0.06488170833333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.194645125,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.06488170833333333,
    "mean_spend_per_solved_task": 0.06488170833333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.06488170833333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 0,
    "solve_rate": 0.0,
    "cap_hits": 3,
    "cap_hit_rate": 1.0,
    "mean_tokens_per_repair_loop": 3226782.3333333335,
    "tokens_per_success": null,
    "conditional_tokens_among_solved_loops": null,
    "mean_tokens_per_solved_task": null,
    "mean_input_tokens_per_repair_loop": 3195868.0,
    "mean_output_tokens_per_repair_loop": 30914.333333333332,
    "mean_cache_read_tokens_per_repair_loop": 2756864.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 19256.0,
    "mean_turns_per_repair_loop": 95.66666666666667,
    "p95_turns_per_repair_loop": 110.0,
    "mean_agent_steps_per_repair_loop": 95.66666666666667,
    "p95_agent_steps_per_repair_loop": 110.0,
    "mean_verifier_submissions_per_repair_loop": 20.0,
    "mean_verifier_submissions_to_success": null,
    "stop_reasons": {
      "verifier_submission_cap": 3
    },
    "total_model_spend_usd": 3.09862179,
    "mean_cost_per_repair_loop": 1.03287393,
    "p95_cost_per_repair_loop": 1.48518842,
    "cpsc": null,
    "realized_cpsc": null,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 3.09862179,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": null,
    "mean_spend_per_solved_task": null,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.99613158,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.03674235000000007,
    "cost_delta_vs_gateway_ratio": 0.03688503681411252
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 13149.0,
    "tokens_per_success": 13149.0,
    "conditional_tokens_among_solved_loops": 13149.0,
    "mean_tokens_per_solved_task": 13149.0,
    "mean_input_tokens_per_repair_loop": 12111.333333333334,
    "mean_output_tokens_per_repair_loop": 1037.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 7040.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 321.6666666666667,
    "mean_turns_per_repair_loop": 8.333333333333334,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.333333333333334,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.03141205,
    "mean_cost_per_repair_loop": 0.010470683333333333,
    "p95_cost_per_repair_loop": 0.01383832,
    "cpsc": 0.010470683333333333,
    "realized_cpsc": 0.010470683333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.03141205,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.010470683333333333,
    "mean_spend_per_solved_task": 0.010470683333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.010470683333333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": -2.3129646346357427e-18,
    "cost_delta_vs_gateway_ratio": -2.2089911049763475e-16
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 588814.3333333334,
    "tokens_per_success": 588814.3333333334,
    "conditional_tokens_among_solved_loops": 588814.3333333334,
    "mean_tokens_per_solved_task": 588814.3333333334,
    "mean_input_tokens_per_repair_loop": 566313.3333333334,
    "mean_output_tokens_per_repair_loop": 22501.0,
    "mean_cache_read_tokens_per_repair_loop": 440533.3333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 12176.666666666666,
    "mean_turns_per_repair_loop": 31.666666666666668,
    "p95_turns_per_repair_loop": 39.0,
    "mean_agent_steps_per_repair_loop": 31.666666666666668,
    "p95_agent_steps_per_repair_loop": 39.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.8753991500000001,
    "mean_cost_per_repair_loop": 0.2917997166666667,
    "p95_cost_per_repair_loop": 0.35762698000000004,
    "cpsc": 0.2917997166666667,
    "realized_cpsc": 0.2917997166666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.8753991500000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.2917997166666667,
    "mean_spend_per_solved_task": 0.2917997166666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.2784722166666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.013327500000000011,
    "cost_delta_vs_gateway_ratio": 0.047859352575747714
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 41459.0,
    "tokens_per_success": 41459.0,
    "conditional_tokens_among_solved_loops": 41459.0,
    "mean_tokens_per_solved_task": 41459.0,
    "mean_input_tokens_per_repair_loop": 38774.0,
    "mean_output_tokens_per_repair_loop": 2685.0,
    "mean_cache_read_tokens_per_repair_loop": 25984.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1406.0,
    "mean_turns_per_repair_loop": 15.666666666666666,
    "p95_turns_per_repair_loop": 23.0,
    "mean_agent_steps_per_repair_loop": 15.666666666666666,
    "p95_agent_steps_per_repair_loop": 23.0,
    "mean_verifier_submissions_per_repair_loop": 2.0,
    "mean_verifier_submissions_to_success": 2.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.08424459,
    "mean_cost_per_repair_loop": 0.028081529999999997,
    "p95_cost_per_repair_loop": 0.048232529999999996,
    "cpsc": 0.028081529999999997,
    "realized_cpsc": 0.028081529999999997,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.08424459,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.028081529999999997,
    "mean_spend_per_solved_task": 0.028081529999999997,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.028081530000000004,
    "mean_cost_delta_vs_gateway_per_repair_loop": -4.625929269271485e-18,
    "cost_delta_vs_gateway_ratio": -1.64732095055771e-16
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 50015.0,
    "tokens_per_success": 50015.0,
    "conditional_tokens_among_solved_loops": 50015.0,
    "mean_tokens_per_solved_task": 50015.0,
    "mean_input_tokens_per_repair_loop": 48474.0,
    "mean_output_tokens_per_repair_loop": 1541.0,
    "mean_cache_read_tokens_per_repair_loop": 34218.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 455.0,
    "mean_turns_per_repair_loop": 16.666666666666668,
    "p95_turns_per_repair_loop": 20.0,
    "mean_agent_steps_per_repair_loop": 16.666666666666668,
    "p95_agent_steps_per_repair_loop": 20.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.07894067,
    "mean_cost_per_repair_loop": 0.026313556666666668,
    "p95_cost_per_repair_loop": 0.03229799,
    "cpsc": 0.026313556666666668,
    "realized_cpsc": 0.026313556666666668,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.07894067,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.026313556666666668,
    "mean_spend_per_solved_task": 0.026313556666666668,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.025878880000000003,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0004346766666666659,
    "cost_delta_vs_gateway_ratio": 0.016796579553159406
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 181998.33333333334,
    "tokens_per_success": 181998.33333333334,
    "conditional_tokens_among_solved_loops": 181998.33333333334,
    "mean_tokens_per_solved_task": 181998.33333333334,
    "mean_input_tokens_per_repair_loop": 177854.66666666666,
    "mean_output_tokens_per_repair_loop": 4143.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 143957.33333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1448.6666666666667,
    "mean_turns_per_repair_loop": 30.333333333333332,
    "p95_turns_per_repair_loop": 36.0,
    "mean_agent_steps_per_repair_loop": 30.333333333333332,
    "p95_agent_steps_per_repair_loop": 36.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.22545579,
    "mean_cost_per_repair_loop": 0.07515192999999999,
    "p95_cost_per_repair_loop": 0.0843714,
    "cpsc": 0.07515192999999999,
    "realized_cpsc": 0.07515192999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.22545579,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07515192999999999,
    "mean_spend_per_solved_task": 0.07515192999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.07371086333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0014410666666666665,
    "cost_delta_vs_gateway_ratio": 0.01955026167784676
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 72722.33333333333,
    "tokens_per_success": 72722.33333333333,
    "conditional_tokens_among_solved_loops": 72722.33333333333,
    "mean_tokens_per_solved_task": 72722.33333333333,
    "mean_input_tokens_per_repair_loop": 69715.66666666667,
    "mean_output_tokens_per_repair_loop": 3006.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 53248.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1473.6666666666667,
    "mean_turns_per_repair_loop": 16.333333333333332,
    "p95_turns_per_repair_loop": 29.0,
    "mean_agent_steps_per_repair_loop": 16.333333333333332,
    "p95_agent_steps_per_repair_loop": 29.0,
    "mean_verifier_submissions_per_repair_loop": 1.6666666666666667,
    "mean_verifier_submissions_to_success": 1.6666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.11309048,
    "mean_cost_per_repair_loop": 0.03769682666666666,
    "p95_cost_per_repair_loop": 0.08512103,
    "cpsc": 0.03769682666666666,
    "realized_cpsc": 0.03769682666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.11309048,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.03769682666666666,
    "mean_spend_per_solved_task": 0.03769682666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03769682666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 43438.666666666664,
    "tokens_per_success": 43438.666666666664,
    "conditional_tokens_among_solved_loops": 43438.666666666664,
    "mean_tokens_per_solved_task": 43438.666666666664,
    "mean_input_tokens_per_repair_loop": 41890.333333333336,
    "mean_output_tokens_per_repair_loop": 1548.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 29354.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 648.0,
    "mean_turns_per_repair_loop": 15.666666666666666,
    "p95_turns_per_repair_loop": 27.0,
    "mean_agent_steps_per_repair_loop": 15.666666666666666,
    "p95_agent_steps_per_repair_loop": 27.0,
    "mean_verifier_submissions_per_repair_loop": 1.6666666666666667,
    "mean_verifier_submissions_to_success": 1.6666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.07139013,
    "mean_cost_per_repair_loop": 0.02379671,
    "p95_cost_per_repair_loop": 0.05081975,
    "cpsc": 0.02379671,
    "realized_cpsc": 0.02379671,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.07139013,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02379671,
    "mean_spend_per_solved_task": 0.02379671,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.022576286666666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0012204233333333314,
    "cost_delta_vs_gateway_ratio": 0.054057753223662616
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 41586.666666666664,
    "tokens_per_success": 41586.666666666664,
    "conditional_tokens_among_solved_loops": 41586.666666666664,
    "mean_tokens_per_solved_task": 41586.666666666664,
    "mean_input_tokens_per_repair_loop": 40029.333333333336,
    "mean_output_tokens_per_repair_loop": 1557.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 25856.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 230.0,
    "mean_turns_per_repair_loop": 15.333333333333334,
    "p95_turns_per_repair_loop": 17.0,
    "mean_agent_steps_per_repair_loop": 15.333333333333334,
    "p95_agent_steps_per_repair_loop": 17.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.07462816,
    "mean_cost_per_repair_loop": 0.024876053333333332,
    "p95_cost_per_repair_loop": 0.03307454,
    "cpsc": 0.024876053333333332,
    "realized_cpsc": 0.024876053333333332,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.07462816,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.024876053333333332,
    "mean_spend_per_solved_task": 0.024876053333333332,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.024876053333333332,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 1,
    "solve_rate": 0.3333333333333333,
    "cap_hits": 2,
    "cap_hit_rate": 0.6666666666666666,
    "mean_tokens_per_repair_loop": 3080704.6666666665,
    "tokens_per_success": 9242114.0,
    "conditional_tokens_among_solved_loops": 1254798.0,
    "mean_tokens_per_solved_task": 1254798.0,
    "mean_input_tokens_per_repair_loop": 3049905.6666666665,
    "mean_output_tokens_per_repair_loop": 30799.0,
    "mean_cache_read_tokens_per_repair_loop": 2823637.3333333335,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 19060.0,
    "mean_turns_per_repair_loop": 86.33333333333333,
    "p95_turns_per_repair_loop": 108.0,
    "mean_agent_steps_per_repair_loop": 86.33333333333333,
    "p95_agent_steps_per_repair_loop": 108.0,
    "mean_verifier_submissions_per_repair_loop": 14.666666666666666,
    "mean_verifier_submissions_to_success": 4.0,
    "stop_reasons": {
      "passed": 1,
      "verifier_submission_cap": 2
    },
    "total_model_spend_usd": 2.49306789,
    "mean_cost_per_repair_loop": 0.83102263,
    "p95_cost_per_repair_loop": 1.12178737,
    "cpsc": 2.49306789,
    "realized_cpsc": 2.49306789,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 2.49306789,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.42063404000000004,
    "mean_spend_per_solved_task": 0.42063404000000004,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.8113352100000001,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0196874199999999,
    "cost_delta_vs_gateway_ratio": 0.024265457430351012
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 38682.333333333336,
    "tokens_per_success": 38682.333333333336,
    "conditional_tokens_among_solved_loops": 38682.333333333336,
    "mean_tokens_per_solved_task": 38682.333333333336,
    "mean_input_tokens_per_repair_loop": 36723.0,
    "mean_output_tokens_per_repair_loop": 1959.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 25642.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 708.6666666666666,
    "mean_turns_per_repair_loop": 14.0,
    "p95_turns_per_repair_loop": 15.0,
    "mean_agent_steps_per_repair_loop": 14.0,
    "p95_agent_steps_per_repair_loop": 15.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.07012466,
    "mean_cost_per_repair_loop": 0.023374886666666667,
    "p95_cost_per_repair_loop": 0.02948636,
    "cpsc": 0.023374886666666667,
    "realized_cpsc": 0.023374886666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.07012466,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.023374886666666667,
    "mean_spend_per_solved_task": 0.023374886666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.02289419,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.000480696666666669,
    "cost_delta_vs_gateway_ratio": 0.020996447861517224
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 275785.0,
    "tokens_per_success": 275785.0,
    "conditional_tokens_among_solved_loops": 275785.0,
    "mean_tokens_per_solved_task": 275785.0,
    "mean_input_tokens_per_repair_loop": 267177.6666666667,
    "mean_output_tokens_per_repair_loop": 8607.333333333334,
    "mean_cache_read_tokens_per_repair_loop": 200021.33333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 6545.333333333333,
    "mean_turns_per_repair_loop": 28.333333333333332,
    "p95_turns_per_repair_loop": 51.0,
    "mean_agent_steps_per_repair_loop": 28.333333333333332,
    "p95_agent_steps_per_repair_loop": 51.0,
    "mean_verifier_submissions_per_repair_loop": 2.3333333333333335,
    "mean_verifier_submissions_to_success": 2.3333333333333335,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.40805898,
    "mean_cost_per_repair_loop": 0.13601966000000001,
    "p95_cost_per_repair_loop": 0.26620606,
    "cpsc": 0.13601966000000001,
    "realized_cpsc": 0.13601966000000001,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.40805898,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.13601966000000001,
    "mean_spend_per_solved_task": 0.13601966000000001,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.1336693733333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.002350286666666701,
    "cost_delta_vs_gateway_ratio": 0.017582835978483688
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 24750.333333333332,
    "tokens_per_success": 24750.333333333332,
    "conditional_tokens_among_solved_loops": 24750.333333333332,
    "mean_tokens_per_solved_task": 24750.333333333332,
    "mean_input_tokens_per_repair_loop": 23811.666666666668,
    "mean_output_tokens_per_repair_loop": 938.6666666666666,
    "mean_cache_read_tokens_per_repair_loop": 17237.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 280.0,
    "mean_turns_per_repair_loop": 13.333333333333334,
    "p95_turns_per_repair_loop": 16.0,
    "mean_agent_steps_per_repair_loop": 13.333333333333334,
    "p95_agent_steps_per_repair_loop": 16.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.03991884,
    "mean_cost_per_repair_loop": 0.013306279999999998,
    "p95_cost_per_repair_loop": 0.01542616,
    "cpsc": 0.013306279999999998,
    "realized_cpsc": 0.013306279999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.03991884,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.013306279999999998,
    "mean_spend_per_solved_task": 0.013306279999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.013306279999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 27609.666666666668,
    "tokens_per_success": 27609.666666666668,
    "conditional_tokens_among_solved_loops": 27609.666666666668,
    "mean_tokens_per_solved_task": 27609.666666666668,
    "mean_input_tokens_per_repair_loop": 26398.333333333332,
    "mean_output_tokens_per_repair_loop": 1211.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 17322.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 432.3333333333333,
    "mean_turns_per_repair_loop": 14.0,
    "p95_turns_per_repair_loop": 16.0,
    "mean_agent_steps_per_repair_loop": 14.0,
    "p95_agent_steps_per_repair_loop": 16.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.05077926,
    "mean_cost_per_repair_loop": 0.01692642,
    "p95_cost_per_repair_loop": 0.024941869999999998,
    "cpsc": 0.01692642,
    "realized_cpsc": 0.01692642,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.05077926,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01692642,
    "mean_spend_per_solved_task": 0.01692642,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.016926419999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 2.3129646346357427e-18,
    "cost_delta_vs_gateway_ratio": 1.3664818872719353e-16
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23526.666666666668,
    "tokens_per_success": 23526.666666666668,
    "conditional_tokens_among_solved_loops": 23526.666666666668,
    "mean_tokens_per_solved_task": 23526.666666666668,
    "mean_input_tokens_per_repair_loop": 22639.0,
    "mean_output_tokens_per_repair_loop": 887.6666666666666,
    "mean_cache_read_tokens_per_repair_loop": 14976.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 305.0,
    "mean_turns_per_repair_loop": 12.666666666666666,
    "p95_turns_per_repair_loop": 13.0,
    "mean_agent_steps_per_repair_loop": 12.666666666666666,
    "p95_agent_steps_per_repair_loop": 13.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.041411909999999996,
    "mean_cost_per_repair_loop": 0.013803969999999999,
    "p95_cost_per_repair_loop": 0.014670369999999999,
    "cpsc": 0.013803969999999999,
    "realized_cpsc": 0.013803969999999999,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.041411909999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.013803969999999999,
    "mean_spend_per_solved_task": 0.013803969999999999,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.013803969999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 488565.6666666667,
    "tokens_per_success": 488565.6666666667,
    "conditional_tokens_among_solved_loops": 488565.6666666667,
    "mean_tokens_per_solved_task": 488565.6666666667,
    "mean_input_tokens_per_repair_loop": 470189.0,
    "mean_output_tokens_per_repair_loop": 18376.666666666668,
    "mean_cache_read_tokens_per_repair_loop": 338816.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 11352.666666666666,
    "mean_turns_per_repair_loop": 30.0,
    "p95_turns_per_repair_loop": 35.0,
    "mean_agent_steps_per_repair_loop": 30.0,
    "p95_agent_steps_per_repair_loop": 35.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.79019166,
    "mean_cost_per_repair_loop": 0.26339722,
    "p95_cost_per_repair_loop": 0.32763364999999994,
    "cpsc": 0.26339722,
    "realized_cpsc": 0.26339722,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.79019166,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.26339722,
    "mean_spend_per_solved_task": 0.26339722,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.26339722,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 34853.333333333336,
    "tokens_per_success": 34853.333333333336,
    "conditional_tokens_among_solved_loops": 34853.333333333336,
    "mean_tokens_per_solved_task": 34853.333333333336,
    "mean_input_tokens_per_repair_loop": 33287.333333333336,
    "mean_output_tokens_per_repair_loop": 1566.0,
    "mean_cache_read_tokens_per_repair_loop": 23381.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 716.3333333333334,
    "mean_turns_per_repair_loop": 14.0,
    "p95_turns_per_repair_loop": 18.0,
    "mean_agent_steps_per_repair_loop": 14.0,
    "p95_agent_steps_per_repair_loop": 18.0,
    "mean_verifier_submissions_per_repair_loop": 1.3333333333333333,
    "mean_verifier_submissions_to_success": 1.3333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06066938,
    "mean_cost_per_repair_loop": 0.020223126666666667,
    "p95_cost_per_repair_loop": 0.03279422,
    "cpsc": 0.020223126666666667,
    "realized_cpsc": 0.020223126666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06066938,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.020223126666666667,
    "mean_spend_per_solved_task": 0.020223126666666667,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.019999393333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0002237333333333346,
    "cost_delta_vs_gateway_ratio": 0.011187006005848908
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 0,
    "solve_rate": 0.0,
    "cap_hits": 3,
    "cap_hit_rate": 1.0,
    "mean_tokens_per_repair_loop": 2744861.3333333335,
    "tokens_per_success": null,
    "conditional_tokens_among_solved_loops": null,
    "mean_tokens_per_solved_task": null,
    "mean_input_tokens_per_repair_loop": 2715379.6666666665,
    "mean_output_tokens_per_repair_loop": 29481.666666666668,
    "mean_cache_read_tokens_per_repair_loop": 2459946.6666666665,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 18283.333333333332,
    "mean_turns_per_repair_loop": 112.0,
    "p95_turns_per_repair_loop": 118.0,
    "mean_agent_steps_per_repair_loop": 112.0,
    "p95_agent_steps_per_repair_loop": 118.0,
    "mean_verifier_submissions_per_repair_loop": 13.333333333333334,
    "mean_verifier_submissions_to_success": null,
    "stop_reasons": {
      "agent_step_cap": 2,
      "verifier_submission_cap": 1
    },
    "total_model_spend_usd": 2.37907405,
    "mean_cost_per_repair_loop": 0.7930246833333333,
    "p95_cost_per_repair_loop": 0.8275796799999999,
    "cpsc": null,
    "realized_cpsc": null,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 2.37907405,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": null,
    "mean_spend_per_solved_task": null,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.7653060533333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.027718629999999855,
    "cost_delta_vs_gateway_ratio": 0.03621901313764579
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 14419.666666666666,
    "tokens_per_success": 14419.666666666666,
    "conditional_tokens_among_solved_loops": 14419.666666666666,
    "mean_tokens_per_solved_task": 14419.666666666666,
    "mean_input_tokens_per_repair_loop": 12741.0,
    "mean_output_tokens_per_repair_loop": 1678.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 9429.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 275.6666666666667,
    "mean_turns_per_repair_loop": 5.333333333333333,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.333333333333333,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06423,
    "mean_cost_per_repair_loop": 0.02141,
    "p95_cost_per_repair_loop": 0.022094,
    "cpsc": 0.02141,
    "realized_cpsc": 0.02141,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06423,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02141,
    "mean_spend_per_solved_task": 0.02141,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.02141,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 20636.666666666668,
    "tokens_per_success": 20636.666666666668,
    "conditional_tokens_among_solved_loops": 20636.666666666668,
    "mean_tokens_per_solved_task": 20636.666666666668,
    "mean_input_tokens_per_repair_loop": 18567.0,
    "mean_output_tokens_per_repair_loop": 2069.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 14208.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 630.3333333333334,
    "mean_turns_per_repair_loop": 7.333333333333333,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.333333333333333,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.08472,
    "mean_cost_per_repair_loop": 0.02824,
    "p95_cost_per_repair_loop": 0.029638,
    "cpsc": 0.02824,
    "realized_cpsc": 0.02824,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.08472,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02824,
    "mean_spend_per_solved_task": 0.02824,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.028239999999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 4.625929269271485e-18,
    "cost_delta_vs_gateway_ratio": 1.638076936710866e-16
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 128751.66666666667,
    "tokens_per_success": 128751.66666666667,
    "conditional_tokens_among_solved_loops": 128751.66666666667,
    "mean_tokens_per_solved_task": 128751.66666666667,
    "mean_input_tokens_per_repair_loop": 112757.0,
    "mean_output_tokens_per_repair_loop": 15994.666666666666,
    "mean_cache_read_tokens_per_repair_loop": 91264.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 2843.6666666666665,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.553758,
    "mean_cost_per_repair_loop": 0.184586,
    "p95_cost_per_repair_loop": 0.202256,
    "cpsc": 0.184586,
    "realized_cpsc": 0.184586,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.553758,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.184586,
    "mean_spend_per_solved_task": 0.184586,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.184586,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 127654.66666666667,
    "tokens_per_success": 127654.66666666667,
    "conditional_tokens_among_solved_loops": 127654.66666666667,
    "mean_tokens_per_solved_task": 127654.66666666667,
    "mean_input_tokens_per_repair_loop": 121984.0,
    "mean_output_tokens_per_repair_loop": 5670.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 107136.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1904.0,
    "mean_turns_per_repair_loop": 18.0,
    "p95_turns_per_repair_loop": 19.0,
    "mean_agent_steps_per_repair_loop": 18.0,
    "p95_agent_steps_per_repair_loop": 19.0,
    "mean_verifier_submissions_per_repair_loop": 2.0,
    "mean_verifier_submissions_to_success": 2.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.351864,
    "mean_cost_per_repair_loop": 0.117288,
    "p95_cost_per_repair_loop": 0.14789,
    "cpsc": 0.117288,
    "realized_cpsc": 0.117288,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.351864,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.117288,
    "mean_spend_per_solved_task": 0.117288,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11728799999999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": 1.5776308810011206e-16
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 19040.666666666668,
    "tokens_per_success": 19040.666666666668,
    "conditional_tokens_among_solved_loops": 19040.666666666668,
    "mean_tokens_per_solved_task": 19040.666666666668,
    "mean_input_tokens_per_repair_loop": 17566.666666666668,
    "mean_output_tokens_per_repair_loop": 1474.0,
    "mean_cache_read_tokens_per_repair_loop": 13397.333333333334,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 270.3333333333333,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.071644,
    "mean_cost_per_repair_loop": 0.023881333333333334,
    "p95_cost_per_repair_loop": 0.025262,
    "cpsc": 0.023881333333333334,
    "realized_cpsc": 0.023881333333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.071644,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.023881333333333334,
    "mean_spend_per_solved_task": 0.023881333333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.023881333333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 36733.0,
    "tokens_per_success": 36733.0,
    "conditional_tokens_among_solved_loops": 36733.0,
    "mean_tokens_per_solved_task": 36733.0,
    "mean_input_tokens_per_repair_loop": 33302.333333333336,
    "mean_output_tokens_per_repair_loop": 3430.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 25002.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 501.6666666666667,
    "mean_turns_per_repair_loop": 6.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 6.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.14905400000000002,
    "mean_cost_per_repair_loop": 0.049684666666666676,
    "p95_cost_per_repair_loop": 0.063186,
    "cpsc": 0.049684666666666676,
    "realized_cpsc": 0.049684666666666676,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.14905400000000002,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.049684666666666676,
    "mean_spend_per_solved_task": 0.049684666666666676,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.049684666666666676,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17515.0,
    "tokens_per_success": 17515.0,
    "conditional_tokens_among_solved_loops": 17515.0,
    "mean_tokens_per_solved_task": 17515.0,
    "mean_input_tokens_per_repair_loop": 16174.666666666666,
    "mean_output_tokens_per_repair_loop": 1340.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 12330.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 279.0,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.065686,
    "mean_cost_per_repair_loop": 0.021895333333333333,
    "p95_cost_per_repair_loop": 0.022812,
    "cpsc": 0.021895333333333333,
    "realized_cpsc": 0.021895333333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.065686,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.021895333333333333,
    "mean_spend_per_solved_task": 0.021895333333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.021895333333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 11881.666666666666,
    "tokens_per_success": 11881.666666666666,
    "conditional_tokens_among_solved_loops": 11881.666666666666,
    "mean_tokens_per_solved_task": 11881.666666666666,
    "mean_input_tokens_per_repair_loop": 10736.0,
    "mean_output_tokens_per_repair_loop": 1145.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 8064.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 417.3333333333333,
    "mean_turns_per_repair_loop": 5.333333333333333,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.333333333333333,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.04875,
    "mean_cost_per_repair_loop": 0.01625,
    "p95_cost_per_repair_loop": 0.020604,
    "cpsc": 0.01625,
    "realized_cpsc": 0.01625,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.04875,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01625,
    "mean_spend_per_solved_task": 0.01625,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.01625,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 24156.333333333332,
    "tokens_per_success": 24156.333333333332,
    "conditional_tokens_among_solved_loops": 24156.333333333332,
    "mean_tokens_per_solved_task": 24156.333333333332,
    "mean_input_tokens_per_repair_loop": 21929.666666666668,
    "mean_output_tokens_per_repair_loop": 2226.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 16725.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 289.6666666666667,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.096394,
    "mean_cost_per_repair_loop": 0.03213133333333333,
    "p95_cost_per_repair_loop": 0.033058,
    "cpsc": 0.03213133333333333,
    "realized_cpsc": 0.03213133333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.096394,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.03213133333333333,
    "mean_spend_per_solved_task": 0.03213133333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03213133333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -4.625929269271485e-18,
    "cost_delta_vs_gateway_ratio": -1.4396941518989207e-16
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 67218.0,
    "tokens_per_success": 67218.0,
    "conditional_tokens_among_solved_loops": 67218.0,
    "mean_tokens_per_solved_task": 67218.0,
    "mean_input_tokens_per_repair_loop": 59756.666666666664,
    "mean_output_tokens_per_repair_loop": 7461.333333333333,
    "mean_cache_read_tokens_per_repair_loop": 46805.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1934.3333333333333,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.28222,
    "mean_cost_per_repair_loop": 0.09407333333333334,
    "p95_cost_per_repair_loop": 0.097236,
    "cpsc": 0.09407333333333334,
    "realized_cpsc": 0.09407333333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.28222,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.09407333333333334,
    "mean_spend_per_solved_task": 0.09407333333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.09407333333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 49382.666666666664,
    "tokens_per_success": 49382.666666666664,
    "conditional_tokens_among_solved_loops": 49382.666666666664,
    "mean_tokens_per_solved_task": 49382.666666666664,
    "mean_input_tokens_per_repair_loop": 43534.0,
    "mean_output_tokens_per_repair_loop": 5848.666666666667,
    "mean_cache_read_tokens_per_repair_loop": 34432.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 2002.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.211536,
    "mean_cost_per_repair_loop": 0.070512,
    "p95_cost_per_repair_loop": 0.08497,
    "cpsc": 0.070512,
    "realized_cpsc": 0.070512,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.211536,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.070512,
    "mean_spend_per_solved_task": 0.070512,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.070512,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 24614.666666666668,
    "tokens_per_success": 24614.666666666668,
    "conditional_tokens_among_solved_loops": 24614.666666666668,
    "mean_tokens_per_solved_task": 24614.666666666668,
    "mean_input_tokens_per_repair_loop": 22050.666666666668,
    "mean_output_tokens_per_repair_loop": 2564.0,
    "mean_cache_read_tokens_per_repair_loop": 16469.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 1173.6666666666667,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.104344,
    "mean_cost_per_repair_loop": 0.03478133333333334,
    "p95_cost_per_repair_loop": 0.046376,
    "cpsc": 0.03478133333333334,
    "realized_cpsc": 0.03478133333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.104344,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.03478133333333334,
    "mean_spend_per_solved_task": 0.03478133333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03478133333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 13652.0,
    "tokens_per_success": 13652.0,
    "conditional_tokens_among_solved_loops": 13652.0,
    "mean_tokens_per_solved_task": 13652.0,
    "mean_input_tokens_per_repair_loop": 12358.0,
    "mean_output_tokens_per_repair_loop": 1294.0,
    "mean_cache_read_tokens_per_repair_loop": 9728.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 370.0,
    "mean_turns_per_repair_loop": 5.666666666666667,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 5.666666666666667,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.053664,
    "mean_cost_per_repair_loop": 0.017888,
    "p95_cost_per_repair_loop": 0.020994,
    "cpsc": 0.017888,
    "realized_cpsc": 0.017888,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.053664,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.017888,
    "mean_spend_per_solved_task": 0.017888,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.017888,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17386.333333333332,
    "tokens_per_success": 17386.333333333332,
    "conditional_tokens_among_solved_loops": 17386.333333333332,
    "mean_tokens_per_solved_task": 17386.333333333332,
    "mean_input_tokens_per_repair_loop": 16257.666666666666,
    "mean_output_tokens_per_repair_loop": 1128.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 12928.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 236.33333333333334,
    "mean_turns_per_repair_loop": 7.0,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 7.0,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.059686,
    "mean_cost_per_repair_loop": 0.019895333333333334,
    "p95_cost_per_repair_loop": 0.021082,
    "cpsc": 0.019895333333333334,
    "realized_cpsc": 0.019895333333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.059686,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.019895333333333334,
    "mean_spend_per_solved_task": 0.019895333333333334,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.019895333333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15131.333333333334,
    "tokens_per_success": 15131.333333333334,
    "conditional_tokens_among_solved_loops": 15131.333333333334,
    "mean_tokens_per_solved_task": 15131.333333333334,
    "mean_input_tokens_per_repair_loop": 13585.0,
    "mean_output_tokens_per_repair_loop": 1546.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 10282.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 262.0,
    "mean_turns_per_repair_loop": 6.0,
    "p95_turns_per_repair_loop": 6.0,
    "mean_agent_steps_per_repair_loop": 6.0,
    "p95_agent_steps_per_repair_loop": 6.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.063072,
    "mean_cost_per_repair_loop": 0.021024,
    "p95_cost_per_repair_loop": 0.022316,
    "cpsc": 0.021024,
    "realized_cpsc": 0.021024,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.063072,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.021024,
    "mean_spend_per_solved_task": 0.021024,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.021023999999999998,
    "mean_cost_delta_vs_gateway_per_repair_loop": 4.625929269271485e-18,
    "cost_delta_vs_gateway_ratio": 2.2003088229031043e-16
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 101626.33333333333,
    "tokens_per_success": 101626.33333333333,
    "conditional_tokens_among_solved_loops": 101626.33333333333,
    "mean_tokens_per_solved_task": 101626.33333333333,
    "mean_input_tokens_per_repair_loop": 92831.66666666667,
    "mean_output_tokens_per_repair_loop": 8794.666666666666,
    "mean_cache_read_tokens_per_repair_loop": 79104.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 3162.6666666666665,
    "mean_turns_per_repair_loop": 10.0,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 10.0,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.359326,
    "mean_cost_per_repair_loop": 0.11977533333333333,
    "p95_cost_per_repair_loop": 0.128706,
    "cpsc": 0.11977533333333333,
    "realized_cpsc": 0.11977533333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.359326,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11977533333333333,
    "mean_spend_per_solved_task": 0.11977533333333333,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.11977533333333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": -1.850371707708594e-17,
    "cost_delta_vs_gateway_ratio": -1.5448687607147222e-16
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 10597.333333333334,
    "tokens_per_success": 10597.333333333334,
    "conditional_tokens_among_solved_loops": 10597.333333333334,
    "mean_tokens_per_solved_task": 10597.333333333334,
    "mean_input_tokens_per_repair_loop": 9530.333333333334,
    "mean_output_tokens_per_repair_loop": 1067.0,
    "mean_cache_read_tokens_per_repair_loop": 7296.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 256.0,
    "mean_turns_per_repair_loop": 4.666666666666667,
    "p95_turns_per_repair_loop": 5.0,
    "mean_agent_steps_per_repair_loop": 4.666666666666667,
    "p95_agent_steps_per_repair_loop": 5.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.043556,
    "mean_cost_per_repair_loop": 0.014518666666666666,
    "p95_cost_per_repair_loop": 0.015732,
    "cpsc": 0.014518666666666666,
    "realized_cpsc": 0.014518666666666666,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.043556,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.014518666666666666,
    "mean_spend_per_solved_task": 0.014518666666666666,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.014518666666666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0,
    "cost_delta_vs_gateway_ratio": 0.0
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 27043.333333333332,
    "tokens_per_success": 27043.333333333332,
    "conditional_tokens_among_solved_loops": 27043.333333333332,
    "mean_tokens_per_solved_task": 27043.333333333332,
    "mean_input_tokens_per_repair_loop": 24615.666666666668,
    "mean_output_tokens_per_repair_loop": 2427.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 19285.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 505.3333333333333,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.104608,
    "mean_cost_per_repair_loop": 0.034869333333333336,
    "p95_cost_per_repair_loop": 0.037632,
    "cpsc": 0.034869333333333336,
    "realized_cpsc": 0.034869333333333336,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.104608,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.034869333333333336,
    "mean_spend_per_solved_task": 0.034869333333333336,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03242933333333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0024400000000000025,
    "cost_delta_vs_gateway_ratio": 0.07524052298330737
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "access-log-to-incidents",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 39533.0,
    "tokens_per_success": 39533.0,
    "conditional_tokens_among_solved_loops": 39533.0,
    "mean_tokens_per_solved_task": 39533.0,
    "mean_input_tokens_per_repair_loop": 36412.333333333336,
    "mean_output_tokens_per_repair_loop": 3120.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 30679.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 11.0,
    "p95_turns_per_repair_loop": 12.0,
    "mean_agent_steps_per_repair_loop": 11.0,
    "p95_agent_steps_per_repair_loop": 12.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06064866000000001,
    "mean_cost_per_repair_loop": 0.020216220000000003,
    "p95_cost_per_repair_loop": 0.02677395,
    "cpsc": 0.020216220000000003,
    "realized_cpsc": 0.020216220000000003,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06064866000000001,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.020216220000000003,
    "mean_spend_per_solved_task": 0.020216220000000003,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.016914103306666664,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.003302116693333338,
    "cost_delta_vs_gateway_ratio": 0.19522859908463577
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "auth-token-expiry-regression",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 37011.0,
    "tokens_per_success": 37011.0,
    "conditional_tokens_among_solved_loops": 37011.0,
    "mean_tokens_per_solved_task": 37011.0,
    "mean_input_tokens_per_repair_loop": 34660.333333333336,
    "mean_output_tokens_per_repair_loop": 2350.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 25181.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06120033,
    "mean_cost_per_repair_loop": 0.02040011,
    "p95_cost_per_repair_loop": 0.02671515,
    "cpsc": 0.02040011,
    "realized_cpsc": 0.02040011,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06120033,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02040011,
    "mean_spend_per_solved_task": 0.02040011,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.018133692133333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0022664178666666666,
    "cost_delta_vs_gateway_ratio": 0.12498380638659567
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "billing-revenue-rollup",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 42951.333333333336,
    "tokens_per_success": 42951.333333333336,
    "conditional_tokens_among_solved_loops": 42951.333333333336,
    "mean_tokens_per_solved_task": 42951.333333333336,
    "mean_input_tokens_per_repair_loop": 39423.0,
    "mean_output_tokens_per_repair_loop": 3528.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 33307.333333333336,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 12.333333333333334,
    "p95_turns_per_repair_loop": 15.0,
    "mean_agent_steps_per_repair_loop": 12.333333333333334,
    "p95_agent_steps_per_repair_loop": 15.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06680367,
    "mean_cost_per_repair_loop": 0.02226789,
    "p95_cost_per_repair_loop": 0.0268614,
    "cpsc": 0.02226789,
    "realized_cpsc": 0.02226789,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06680367,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02226789,
    "mean_spend_per_solved_task": 0.02226789,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.01845877237333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.003809117626666666,
    "cost_delta_vs_gateway_ratio": 0.20635812336954484
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "cache-invalidates-on-settings-change",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 55524.333333333336,
    "tokens_per_success": 55524.333333333336,
    "conditional_tokens_among_solved_loops": 55524.333333333336,
    "mean_tokens_per_solved_task": 55524.333333333336,
    "mean_input_tokens_per_repair_loop": 52039.666666666664,
    "mean_output_tokens_per_repair_loop": 3484.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 45849.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 14.666666666666666,
    "p95_turns_per_repair_loop": 19.0,
    "mean_agent_steps_per_repair_loop": 14.666666666666666,
    "p95_agent_steps_per_repair_loop": 19.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.07339242,
    "mean_cost_per_repair_loop": 0.02446414,
    "p95_cost_per_repair_loop": 0.03751455,
    "cpsc": 0.02446414,
    "realized_cpsc": 0.02446414,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.07339242,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02446414,
    "mean_spend_per_solved_task": 0.02446414,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.022591711626666663,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0018724283733333365,
    "cost_delta_vs_gateway_ratio": 0.08288120901486593
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "config-flag-ignored",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 30972.333333333332,
    "tokens_per_success": 30972.333333333332,
    "conditional_tokens_among_solved_loops": 30972.333333333332,
    "mean_tokens_per_solved_task": 30972.333333333332,
    "mean_input_tokens_per_repair_loop": 29234.666666666668,
    "mean_output_tokens_per_repair_loop": 1737.6666666666667,
    "mean_cache_read_tokens_per_repair_loop": 22160.666666666668,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.04734222,
    "mean_cost_per_repair_loop": 0.015780739999999998,
    "p95_cost_per_repair_loop": 0.01986495,
    "cpsc": 0.015780739999999998,
    "realized_cpsc": 0.015780739999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.04734222,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.015780739999999998,
    "mean_spend_per_solved_task": 0.015780739999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.014281070279999997,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0014996697200000016,
    "cost_delta_vs_gateway_ratio": 0.10501101742354858
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "config-key-rollover",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 70320.66666666667,
    "tokens_per_success": 70320.66666666667,
    "conditional_tokens_among_solved_loops": 70320.66666666667,
    "mean_tokens_per_solved_task": 70320.66666666667,
    "mean_input_tokens_per_repair_loop": 66625.0,
    "mean_output_tokens_per_repair_loop": 3695.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 56599.666666666664,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 14.0,
    "p95_turns_per_repair_loop": 18.0,
    "mean_agent_steps_per_repair_loop": 14.0,
    "p95_agent_steps_per_repair_loop": 18.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0917955,
    "mean_cost_per_repair_loop": 0.0305985,
    "p95_cost_per_repair_loop": 0.03393816,
    "cpsc": 0.0305985,
    "realized_cpsc": 0.0305985,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0917955,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0305985,
    "mean_spend_per_solved_task": 0.0305985,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.028347805093333334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.002250694906666665,
    "cost_delta_vs_gateway_ratio": 0.07939573802121173
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "env-flags-to-json",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 30254.666666666668,
    "tokens_per_success": 30254.666666666668,
    "conditional_tokens_among_solved_loops": 30254.666666666668,
    "mean_tokens_per_solved_task": 30254.666666666668,
    "mean_input_tokens_per_repair_loop": 27887.666666666668,
    "mean_output_tokens_per_repair_loop": 2367.0,
    "mean_cache_read_tokens_per_repair_loop": 23268.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 10.0,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 10.0,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.04675584,
    "mean_cost_per_repair_loop": 0.01558528,
    "p95_cost_per_repair_loop": 0.01799856,
    "cpsc": 0.01558528,
    "realized_cpsc": 0.01558528,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.04675584,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.01558528,
    "mean_spend_per_solved_task": 0.01558528,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.013502393653333333,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0020828863466666667,
    "cost_delta_vs_gateway_ratio": 0.15426052595885212
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "extract-error-fields",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 20522.333333333332,
    "tokens_per_success": 20522.333333333332,
    "conditional_tokens_among_solved_loops": 20522.333333333332,
    "mean_tokens_per_solved_task": 20522.333333333332,
    "mean_input_tokens_per_repair_loop": 19220.0,
    "mean_output_tokens_per_repair_loop": 1302.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 16214.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 11.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 11.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0288633,
    "mean_cost_per_repair_loop": 0.0096211,
    "p95_cost_per_repair_loop": 0.01244175,
    "cpsc": 0.0096211,
    "realized_cpsc": 0.0096211,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0288633,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0096211,
    "mean_spend_per_solved_task": 0.0096211,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.00786324628,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0017578537200000008,
    "cost_delta_vs_gateway_ratio": 0.22355318114238193
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "invoice-cli-regression-test-fix",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 29736.333333333332,
    "tokens_per_success": 29736.333333333332,
    "conditional_tokens_among_solved_loops": 29736.333333333332,
    "mean_tokens_per_solved_task": 29736.333333333332,
    "mean_input_tokens_per_repair_loop": 28465.0,
    "mean_output_tokens_per_repair_loop": 1271.3333333333333,
    "mean_cache_read_tokens_per_repair_loop": 21970.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 12.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 12.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.0414261,
    "mean_cost_per_repair_loop": 0.0138087,
    "p95_cost_per_repair_loop": 0.01569489,
    "cpsc": 0.0138087,
    "realized_cpsc": 0.0138087,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.0414261,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0138087,
    "mean_spend_per_solved_task": 0.0138087,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.011638978706666666,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.002169721293333334,
    "cost_delta_vs_gateway_ratio": 0.18641852932427344
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "invoice-multi-source-merge",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 2,
    "solve_rate": 0.6666666666666666,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 117213.33333333333,
    "tokens_per_success": 175820.0,
    "conditional_tokens_among_solved_loops": 155677.0,
    "mean_tokens_per_solved_task": 155677.0,
    "mean_input_tokens_per_repair_loop": 96651.0,
    "mean_output_tokens_per_repair_loop": 20562.333333333332,
    "mean_cache_read_tokens_per_repair_loop": 73087.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 0.6666666666666666,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "agent_exit_repeatedformaterror": 1,
      "passed": 2
    },
    "total_model_spend_usd": 0.29027154,
    "mean_cost_per_repair_loop": 0.09675718,
    "p95_cost_per_repair_loop": 0.11825088,
    "cpsc": 0.14513577,
    "realized_cpsc": 0.14513577,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.29027154,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.10307811,
    "mean_spend_per_solved_task": 0.10307811,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04365391752,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.053103262479999996,
    "cost_delta_vs_gateway_ratio": 1.2164604117298472
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "merge-divergent-config-branches",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 32735.333333333332,
    "tokens_per_success": 32735.333333333332,
    "conditional_tokens_among_solved_loops": 32735.333333333332,
    "mean_tokens_per_solved_task": 32735.333333333332,
    "mean_input_tokens_per_repair_loop": 28740.666666666668,
    "mean_output_tokens_per_repair_loop": 3994.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 22274.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06602121,
    "mean_cost_per_repair_loop": 0.02200707,
    "p95_cost_per_repair_loop": 0.02455929,
    "cpsc": 0.02200707,
    "realized_cpsc": 0.02200707,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06602121,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02200707,
    "mean_spend_per_solved_task": 0.02200707,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.02105069310666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0009563768933333281,
    "cost_delta_vs_gateway_ratio": 0.045432085703175605
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "payout-reconcile",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 60004.333333333336,
    "tokens_per_success": 60004.333333333336,
    "conditional_tokens_among_solved_loops": 60004.333333333336,
    "mean_tokens_per_solved_task": 60004.333333333336,
    "mean_input_tokens_per_repair_loop": 52908.333333333336,
    "mean_output_tokens_per_repair_loop": 7096.0,
    "mean_cache_read_tokens_per_repair_loop": 43673.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.666666666666666,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 8.666666666666666,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.11321399999999998,
    "mean_cost_per_repair_loop": 0.037737999999999994,
    "p95_cost_per_repair_loop": 0.04379643,
    "cpsc": 0.037737999999999994,
    "realized_cpsc": 0.037737999999999994,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.11321399999999998,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.037737999999999994,
    "mean_spend_per_solved_task": 0.037737999999999994,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.03109600222666667,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.006641997773333325,
    "cost_delta_vs_gateway_ratio": 0.21359651716378567
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "post-build-status",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 17309.666666666668,
    "tokens_per_success": 17309.666666666668,
    "conditional_tokens_among_solved_loops": 17309.666666666668,
    "mean_tokens_per_solved_task": 17309.666666666668,
    "mean_input_tokens_per_repair_loop": 16187.666666666666,
    "mean_output_tokens_per_repair_loop": 1122.0,
    "mean_cache_read_tokens_per_repair_loop": 12841.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.666666666666667,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 7.666666666666667,
    "p95_agent_steps_per_repair_loop": 9.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.02636784,
    "mean_cost_per_repair_loop": 0.00878928,
    "p95_cost_per_repair_loop": 0.0102633,
    "cpsc": 0.00878928,
    "realized_cpsc": 0.00878928,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.02636784,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.00878928,
    "mean_spend_per_solved_task": 0.00878928,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.008415120866666665,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.000374159133333334,
    "cost_delta_vs_gateway_ratio": 0.04446271649114686
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "retry-error-fallback",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 50830.333333333336,
    "tokens_per_success": 50830.333333333336,
    "conditional_tokens_among_solved_loops": 50830.333333333336,
    "mean_tokens_per_solved_task": 50830.333333333336,
    "mean_input_tokens_per_repair_loop": 48381.0,
    "mean_output_tokens_per_repair_loop": 2449.3333333333335,
    "mean_cache_read_tokens_per_repair_loop": 42207.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 14.333333333333334,
    "p95_turns_per_repair_loop": 24.0,
    "mean_agent_steps_per_repair_loop": 14.333333333333334,
    "p95_agent_steps_per_repair_loop": 24.0,
    "mean_verifier_submissions_per_repair_loop": 2.0,
    "mean_verifier_submissions_to_success": 2.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.06206124,
    "mean_cost_per_repair_loop": 0.02068708,
    "p95_cost_per_repair_loop": 0.04087203,
    "cpsc": 0.02068708,
    "realized_cpsc": 0.02068708,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.06206124,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02068708,
    "mean_spend_per_solved_task": 0.02068708,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.0188522334,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0018348465999999976,
    "cost_delta_vs_gateway_ratio": 0.09732781050758672
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "split-notification-renderer",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 23314.666666666668,
    "tokens_per_success": 23314.666666666668,
    "conditional_tokens_among_solved_loops": 23314.666666666668,
    "mean_tokens_per_solved_task": 23314.666666666668,
    "mean_input_tokens_per_repair_loop": 21556.666666666668,
    "mean_output_tokens_per_repair_loop": 1758.0,
    "mean_cache_read_tokens_per_repair_loop": 17844.333333333332,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.666666666666666,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 9.666666666666666,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.035815349999999996,
    "mean_cost_per_repair_loop": 0.011938449999999998,
    "p95_cost_per_repair_loop": 0.01407999,
    "cpsc": 0.011938449999999998,
    "realized_cpsc": 0.011938449999999998,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.035815349999999996,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.011938449999999998,
    "mean_spend_per_solved_task": 0.011938449999999998,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.010350380746666665,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0015880692533333334,
    "cost_delta_vs_gateway_ratio": 0.15343099854995867
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "support-sla-business-hours",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 109971.66666666667,
    "tokens_per_success": 109971.66666666667,
    "conditional_tokens_among_solved_loops": 109971.66666666667,
    "mean_tokens_per_solved_task": 109971.66666666667,
    "mean_input_tokens_per_repair_loop": 101955.66666666667,
    "mean_output_tokens_per_repair_loop": 8016.0,
    "mean_cache_read_tokens_per_repair_loop": 76964.33333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 13.333333333333334,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 13.333333333333334,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.0,
    "mean_verifier_submissions_to_success": 1.0,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.18343056,
    "mean_cost_per_repair_loop": 0.06114352,
    "p95_cost_per_repair_loop": 0.07371927,
    "cpsc": 0.06114352,
    "realized_cpsc": 0.06114352,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.18343056,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.06114352,
    "mean_spend_per_solved_task": 0.06114352,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.04889466732000001,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.012248852679999986,
    "cost_delta_vs_gateway_ratio": 0.2505151042307978
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "ticket-cut-from-bug-report",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 34607.333333333336,
    "tokens_per_success": 34607.333333333336,
    "conditional_tokens_among_solved_loops": 34607.333333333336,
    "mean_tokens_per_solved_task": 34607.333333333336,
    "mean_input_tokens_per_repair_loop": 31589.666666666668,
    "mean_output_tokens_per_repair_loop": 3017.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 24432.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.333333333333334,
    "p95_turns_per_repair_loop": 16.0,
    "mean_agent_steps_per_repair_loop": 9.333333333333334,
    "p95_agent_steps_per_repair_loop": 16.0,
    "mean_verifier_submissions_per_repair_loop": 1.3333333333333333,
    "mean_verifier_submissions_to_success": 1.3333333333333333,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.060322169999999994,
    "mean_cost_per_repair_loop": 0.02010739,
    "p95_cost_per_repair_loop": 0.038508089999999995,
    "cpsc": 0.02010739,
    "realized_cpsc": 0.02010739,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.060322169999999994,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.02010739,
    "mean_spend_per_solved_task": 0.02010739,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.018777132413333335,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0013302575866666634,
    "cost_delta_vs_gateway_ratio": 0.07084455482254944
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "task_id": "ticket-update-dont-duplicate",
    "total_trials": 3,
    "repair_loops": 3,
    "excluded_repair_loops": 0,
    "successes": 3,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 141876.66666666666,
    "tokens_per_success": 141876.66666666666,
    "conditional_tokens_among_solved_loops": 141876.66666666666,
    "mean_tokens_per_solved_task": 141876.66666666666,
    "mean_input_tokens_per_repair_loop": 133470.66666666666,
    "mean_output_tokens_per_repair_loop": 8406.0,
    "mean_cache_read_tokens_per_repair_loop": 117834.33333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 18.666666666666668,
    "p95_turns_per_repair_loop": 35.0,
    "mean_agent_steps_per_repair_loop": 18.666666666666668,
    "p95_agent_steps_per_repair_loop": 35.0,
    "mean_verifier_submissions_per_repair_loop": 1.6666666666666667,
    "mean_verifier_submissions_to_success": 1.6666666666666667,
    "stop_reasons": {
      "passed": 3
    },
    "total_model_spend_usd": 0.18290991,
    "mean_cost_per_repair_loop": 0.060969970000000005,
    "p95_cost_per_repair_loop": 0.14758944000000002,
    "cpsc": 0.060969970000000005,
    "realized_cpsc": 0.060969970000000005,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 0.18290991,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.060969970000000005,
    "mean_spend_per_solved_task": 0.060969970000000005,
    "gateway_reported_repair_loops": 3,
    "mean_gateway_reported_cost_per_repair_loop": 0.05730378283999999,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0036661871600000098,
    "cost_delta_vs_gateway_ratio": 0.06397810019342888
  }
]
