[
  {
    "model_config": "openrouter/anthropic/claude-fable-5[low]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 15342.833333333334,
    "tokens_per_success": 15342.833333333334,
    "conditional_tokens_among_solved_loops": 15342.833333333334,
    "mean_tokens_per_solved_task": 15342.833333333334,
    "mean_input_tokens_per_repair_loop": 14076.888888888889,
    "mean_output_tokens_per_repair_loop": 1265.9444444444443,
    "mean_cache_read_tokens_per_repair_loop": 9607.925925925925,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 4.62962962962963,
    "p95_turns_per_repair_loop": 7.0,
    "mean_agent_steps_per_repair_loop": 4.62962962962963,
    "p95_agent_steps_per_repair_loop": 7.0,
    "mean_verifier_submissions_per_repair_loop": 1.037037037037037,
    "mean_verifier_submissions_to_success": 1.037037037037037,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 6.350118,
    "mean_cost_per_repair_loop": 0.11759477777777778,
    "p95_cost_per_repair_loop": 0.326736,
    "cpsc": 0.11759477777777778,
    "realized_cpsc": 0.11759477777777778,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 6.350118,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.11759477777777778,
    "mean_spend_per_solved_task": 0.11759477777777778,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.12874403703703705,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.011149259259259256,
    "cost_delta_vs_gateway_ratio": -0.08660019924691224
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[low]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 24816.444444444445,
    "tokens_per_success": 24816.444444444445,
    "conditional_tokens_among_solved_loops": 24816.444444444445,
    "mean_tokens_per_solved_task": 24816.444444444445,
    "mean_input_tokens_per_repair_loop": 23081.703703703704,
    "mean_output_tokens_per_repair_loop": 1734.7407407407406,
    "mean_cache_read_tokens_per_repair_loop": 17497.962962962964,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 5.962962962962963,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 5.962962962962963,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.0925925925925926,
    "mean_verifier_submissions_to_success": 1.0925925925925926,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 4.321955,
    "mean_cost_per_repair_loop": 0.0800362037037037,
    "p95_cost_per_repair_loop": 0.1950455,
    "cpsc": 0.0800362037037037,
    "realized_cpsc": 0.0800362037037037,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 4.321955,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.0800362037037037,
    "mean_spend_per_solved_task": 0.0800362037037037,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.08700097222222222,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.006964768518518522,
    "cost_delta_vs_gateway_ratio": -0.08005391595772933
  },
  {
    "model_config": "openrouter/anthropic/claude-opus-4.8[medium]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 38642.851851851854,
    "tokens_per_success": 38642.851851851854,
    "conditional_tokens_among_solved_loops": 38642.851851851854,
    "mean_tokens_per_solved_task": 38642.851851851854,
    "mean_input_tokens_per_repair_loop": 35743.48148148148,
    "mean_output_tokens_per_repair_loop": 2899.3703703703704,
    "mean_cache_read_tokens_per_repair_loop": 28233.90740740741,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.5,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 7.5,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.0185185185185186,
    "mean_verifier_submissions_to_success": 1.0185185185185186,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 6.7040505,
    "mean_cost_per_repair_loop": 0.12414908333333334,
    "p95_cost_per_repair_loop": 0.3171925,
    "cpsc": 0.12414908333333334,
    "realized_cpsc": 0.12414908333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 6.7040505,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.12414908333333334,
    "mean_spend_per_solved_task": 0.12414908333333334,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.13351730092592592,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.009368217592592588,
    "cost_delta_vs_gateway_ratio": -0.07016482154466246
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[low]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 67624.85185185185,
    "tokens_per_success": 67624.85185185185,
    "conditional_tokens_among_solved_loops": 67624.85185185185,
    "mean_tokens_per_solved_task": 67624.85185185185,
    "mean_input_tokens_per_repair_loop": 65288.7037037037,
    "mean_output_tokens_per_repair_loop": 2336.1481481481483,
    "mean_cache_read_tokens_per_repair_loop": 58230.555555555555,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 8.518518518518519,
    "p95_turns_per_repair_loop": 18.0,
    "mean_agent_steps_per_repair_loop": 8.518518518518519,
    "p95_agent_steps_per_repair_loop": 18.0,
    "mean_verifier_submissions_per_repair_loop": 1.6481481481481481,
    "mean_verifier_submissions_to_success": 1.6481481481481481,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 3.979035,
    "mean_cost_per_repair_loop": 0.07368583333333334,
    "p95_cost_per_repair_loop": 0.1609584,
    "cpsc": 0.07368583333333334,
    "realized_cpsc": 0.07368583333333334,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 3.979035,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07368583333333334,
    "mean_spend_per_solved_task": 0.07368583333333334,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.052644444444444445,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.02104138888888889,
    "cost_delta_vs_gateway_ratio": 0.3996886872097932
  },
  {
    "model_config": "openrouter/anthropic/claude-sonnet-5[medium]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 47832.07407407407,
    "tokens_per_success": 47832.07407407407,
    "conditional_tokens_among_solved_loops": 47832.07407407407,
    "mean_tokens_per_solved_task": 47832.07407407407,
    "mean_input_tokens_per_repair_loop": 45099.40740740741,
    "mean_output_tokens_per_repair_loop": 2732.6666666666665,
    "mean_cache_read_tokens_per_repair_loop": 35097.81481481482,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.185185185185185,
    "p95_turns_per_repair_loop": 15.0,
    "mean_agent_steps_per_repair_loop": 9.185185185185185,
    "p95_agent_steps_per_repair_loop": 15.0,
    "mean_verifier_submissions_per_repair_loop": 1.037037037037037,
    "mean_verifier_submissions_to_success": 1.037037037037037,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 4.4023026,
    "mean_cost_per_repair_loop": 0.08152412222222222,
    "p95_cost_per_repair_loop": 0.2018985,
    "cpsc": 0.08152412222222222,
    "realized_cpsc": 0.08152412222222222,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 4.4023026,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.08152412222222222,
    "mean_spend_per_solved_task": 0.08152412222222222,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.059341025925925925,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.022183096296296292,
    "cost_delta_vs_gateway_ratio": 0.3738239430505795
  },
  {
    "model_config": "openrouter/google/gemini-3.5-flash[medium]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 307752.6296296296,
    "tokens_per_success": 307752.6296296296,
    "conditional_tokens_among_solved_loops": 307752.6296296296,
    "mean_tokens_per_solved_task": 307752.6296296296,
    "mean_input_tokens_per_repair_loop": 290188.8333333333,
    "mean_output_tokens_per_repair_loop": 17563.796296296296,
    "mean_cache_read_tokens_per_repair_loop": 198208.55555555556,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 24.90740740740741,
    "p95_turns_per_repair_loop": 49.0,
    "mean_agent_steps_per_repair_loop": 24.90740740740741,
    "p95_agent_steps_per_repair_loop": 49.0,
    "mean_verifier_submissions_per_repair_loop": 1.2037037037037037,
    "mean_verifier_submissions_to_success": 1.2037037037037037,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 17.5918968,
    "mean_cost_per_repair_loop": 0.3257758666666667,
    "p95_cost_per_repair_loop": 1.39996725,
    "cpsc": 0.3257758666666667,
    "realized_cpsc": 0.3257758666666667,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 17.5918968,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.3257758666666667,
    "mean_spend_per_solved_task": 0.3257758666666667,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.30927439444444443,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.016501472222222223,
    "cost_delta_vs_gateway_ratio": 0.053355442670461405
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k2.7-code",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 117502.38888888889,
    "tokens_per_success": 117502.38888888889,
    "conditional_tokens_among_solved_loops": 117502.38888888889,
    "mean_tokens_per_solved_task": 117502.38888888889,
    "mean_input_tokens_per_repair_loop": 111682.42592592593,
    "mean_output_tokens_per_repair_loop": 5819.962962962963,
    "mean_cache_read_tokens_per_repair_loop": 91205.33333333333,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 14.148148148148149,
    "p95_turns_per_repair_loop": 28.0,
    "mean_agent_steps_per_repair_loop": 14.148148148148149,
    "p95_agent_steps_per_repair_loop": 28.0,
    "mean_verifier_submissions_per_repair_loop": 1.2407407407407407,
    "mean_verifier_submissions_to_success": 1.2407407407407407,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 2.65700082,
    "mean_cost_per_repair_loop": 0.04920371888888889,
    "p95_cost_per_repair_loop": 0.22189722,
    "cpsc": 0.04920371888888889,
    "realized_cpsc": 0.04920371888888889,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 2.65700082,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.04920371888888889,
    "mean_spend_per_solved_task": 0.04920371888888889,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.054128717629629626,
    "mean_cost_delta_vs_gateway_per_repair_loop": -0.004924998740740738,
    "cost_delta_vs_gateway_ratio": -0.09098679880871283
  },
  {
    "model_config": "openrouter/moonshotai/kimi-k3[max]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 353648.72222222225,
    "tokens_per_success": 353648.72222222225,
    "conditional_tokens_among_solved_loops": 353648.72222222225,
    "mean_tokens_per_solved_task": 353648.72222222225,
    "mean_input_tokens_per_repair_loop": 339439.8703703704,
    "mean_output_tokens_per_repair_loop": 14208.851851851852,
    "mean_cache_read_tokens_per_repair_loop": 317098.6666666667,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 9019.203703703704,
    "mean_turns_per_repair_loop": 15.351851851851851,
    "p95_turns_per_repair_loop": 62.0,
    "mean_agent_steps_per_repair_loop": 15.351851851851851,
    "p95_agent_steps_per_repair_loop": 62.0,
    "mean_verifier_submissions_per_repair_loop": 1.537037037037037,
    "mean_verifier_submissions_to_success": 1.537037037037037,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 20.2654434,
    "mean_cost_per_repair_loop": 0.37528598888888887,
    "p95_cost_per_repair_loop": 2.1540828,
    "cpsc": 0.37528598888888887,
    "realized_cpsc": 0.37528598888888887,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 20.2654434,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.37528598888888887,
    "mean_spend_per_solved_task": 0.37528598888888887,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.34628535555555556,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.02900063333333328,
    "cost_delta_vs_gateway_ratio": 0.08374779027778038
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[low]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 26019.925925925927,
    "tokens_per_success": 26019.925925925927,
    "conditional_tokens_among_solved_loops": 26019.925925925927,
    "mean_tokens_per_solved_task": 26019.925925925927,
    "mean_input_tokens_per_repair_loop": 23711.314814814814,
    "mean_output_tokens_per_repair_loop": 2308.6111111111113,
    "mean_cache_read_tokens_per_repair_loop": 12202.666666666666,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 7.5,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 7.5,
    "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": 54
    },
    "total_model_spend_usd": 7.176757,
    "mean_cost_per_repair_loop": 0.1329029074074074,
    "p95_cost_per_repair_loop": 0.223198,
    "cpsc": 0.1329029074074074,
    "realized_cpsc": 0.1329029074074074,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 7.176757,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.1329029074074074,
    "mean_spend_per_solved_task": 0.1329029074074074,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.13215687037037038,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0007460370370370379,
    "cost_delta_vs_gateway_ratio": 0.005645087046524822
  },
  {
    "model_config": "openrouter/openai/gpt-5.5[medium]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 61552.59259259259,
    "tokens_per_success": 61552.59259259259,
    "conditional_tokens_among_solved_loops": 61552.59259259259,
    "mean_tokens_per_solved_task": 61552.59259259259,
    "mean_input_tokens_per_repair_loop": 57491.648148148146,
    "mean_output_tokens_per_repair_loop": 4060.9444444444443,
    "mean_cache_read_tokens_per_repair_loop": 39168.0,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 9.277777777777779,
    "p95_turns_per_repair_loop": 14.0,
    "mean_agent_steps_per_repair_loop": 9.277777777777779,
    "p95_agent_steps_per_repair_loop": 14.0,
    "mean_verifier_submissions_per_repair_loop": 1.1296296296296295,
    "mean_verifier_submissions_to_success": 1.1296296296296295,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 12.583651,
    "mean_cost_per_repair_loop": 0.23303057407407407,
    "p95_cost_per_repair_loop": 0.370568,
    "cpsc": 0.23303057407407407,
    "realized_cpsc": 0.23303057407407407,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 12.583651,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.23303057407407407,
    "mean_spend_per_solved_task": 0.23303057407407407,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.23303057407407407,
    "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]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 53,
    "solve_rate": 0.9814814814814815,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 22369.685185185186,
    "tokens_per_success": 22791.754716981133,
    "conditional_tokens_among_solved_loops": 20416.924528301886,
    "mean_tokens_per_solved_task": 20416.924528301886,
    "mean_input_tokens_per_repair_loop": 20254.314814814814,
    "mean_output_tokens_per_repair_loop": 2115.3703703703704,
    "mean_cache_read_tokens_per_repair_loop": 14975.24074074074,
    "mean_cache_write_tokens_per_repair_loop": 4485.722222222223,
    "mean_reasoning_tokens_per_repair_loop": 306.44444444444446,
    "mean_turns_per_repair_loop": 6.2592592592592595,
    "p95_turns_per_repair_loop": 8.0,
    "mean_agent_steps_per_repair_loop": 6.2592592592592595,
    "p95_agent_steps_per_repair_loop": 8.0,
    "mean_verifier_submissions_per_repair_loop": 1.1111111111111112,
    "mean_verifier_submissions_to_success": 1.0188679245283019,
    "stop_reasons": {
      "agent_exit_repeatedformaterror": 1,
      "passed": 53
    },
    "total_model_spend_usd": 1.11187355,
    "mean_cost_per_repair_loop": 0.020590250925925927,
    "p95_cost_per_repair_loop": 0.036359050000000004,
    "cpsc": 0.020978746226415095,
    "realized_cpsc": 0.020978746226415095,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.11187355,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.019982904716981133,
    "mean_spend_per_solved_task": 0.019982904716981133,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.02008630648148148,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.0005039444444444473,
    "cost_delta_vs_gateway_ratio": 0.025088955249640225
  },
  {
    "model_config": "openrouter/openai/gpt-5.6-sol[low]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 28640.87037037037,
    "tokens_per_success": 28640.87037037037,
    "conditional_tokens_among_solved_loops": 28640.87037037037,
    "mean_tokens_per_solved_task": 28640.87037037037,
    "mean_input_tokens_per_repair_loop": 25403.12962962963,
    "mean_output_tokens_per_repair_loop": 3237.740740740741,
    "mean_cache_read_tokens_per_repair_loop": 18137.40740740741,
    "mean_cache_write_tokens_per_repair_loop": 6434.240740740741,
    "mean_reasoning_tokens_per_repair_loop": 459.5,
    "mean_turns_per_repair_loop": 6.796296296296297,
    "p95_turns_per_repair_loop": 9.0,
    "mean_agent_steps_per_repair_loop": 6.796296296296297,
    "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": 54
    },
    "total_model_spend_usd": 8.13090625,
    "mean_cost_per_repair_loop": 0.150572337962963,
    "p95_cost_per_repair_loop": 0.25486475,
    "cpsc": 0.150572337962963,
    "realized_cpsc": 0.150572337962963,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 8.13090625,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.150572337962963,
    "mean_spend_per_solved_task": 0.150572337962963,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.150572337962963,
    "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]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 53,
    "solve_rate": 0.9814814814814815,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 32534.185185185186,
    "tokens_per_success": 33148.03773584906,
    "conditional_tokens_among_solved_loops": 28962.245283018867,
    "mean_tokens_per_solved_task": 28962.245283018867,
    "mean_input_tokens_per_repair_loop": 28993.98148148148,
    "mean_output_tokens_per_repair_loop": 3540.203703703704,
    "mean_cache_read_tokens_per_repair_loop": 21668.90740740741,
    "mean_cache_write_tokens_per_repair_loop": 6276.444444444444,
    "mean_reasoning_tokens_per_repair_loop": 777.4629629629629,
    "mean_turns_per_repair_loop": 7.111111111111111,
    "p95_turns_per_repair_loop": 10.0,
    "mean_agent_steps_per_repair_loop": 7.111111111111111,
    "p95_agent_steps_per_repair_loop": 10.0,
    "mean_verifier_submissions_per_repair_loop": 1.1296296296296295,
    "mean_verifier_submissions_to_success": 1.0377358490566038,
    "stop_reasons": {
      "agent_exit_repeatedformaterror": 1,
      "passed": 53
    },
    "total_model_spend_usd": 4.36081025,
    "mean_cost_per_repair_loop": 0.08075574537037038,
    "p95_cost_per_repair_loop": 0.147829125,
    "cpsc": 0.08227943867924528,
    "realized_cpsc": 0.08227943867924528,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 4.36081025,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07793531839622642,
    "mean_spend_per_solved_task": 0.07793531839622642,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.07916358101851852,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.001592164351851853,
    "cost_delta_vs_gateway_ratio": 0.020112333618149517
  },
  {
    "model_config": "openrouter/thinkingmachines/inkling[high]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 46,
    "solve_rate": 0.8518518518518519,
    "cap_hits": 8,
    "cap_hit_rate": 0.14814814814814814,
    "mean_tokens_per_repair_loop": 611072.4814814815,
    "tokens_per_success": 717345.9565217391,
    "conditional_tokens_among_solved_loops": 154253.63043478262,
    "mean_tokens_per_solved_task": 154253.63043478262,
    "mean_input_tokens_per_repair_loop": 602030.1111111111,
    "mean_output_tokens_per_repair_loop": 9042.37037037037,
    "mean_cache_read_tokens_per_repair_loop": 524335.4074074074,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 5283.2962962962965,
    "mean_turns_per_repair_loop": 31.685185185185187,
    "p95_turns_per_repair_loop": 110.0,
    "mean_agent_steps_per_repair_loop": 31.685185185185187,
    "p95_agent_steps_per_repair_loop": 110.0,
    "mean_verifier_submissions_per_repair_loop": 3.7222222222222223,
    "mean_verifier_submissions_to_success": 1.326086956521739,
    "stop_reasons": {
      "agent_step_cap": 2,
      "passed": 46,
      "verifier_submission_cap": 6
    },
    "total_model_spend_usd": 10.98647944,
    "mean_cost_per_repair_loop": 0.20345332296296295,
    "p95_cost_per_repair_loop": 0.95064648,
    "cpsc": 0.2388365095652174,
    "realized_cpsc": 0.2388365095652174,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 10.98647944,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.07470325543478261,
    "mean_spend_per_solved_task": 0.07470325543478261,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.19769627944444446,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.005757043518518513,
    "cost_delta_vs_gateway_ratio": 0.02912064675519767
  },
  {
    "model_config": "openrouter/x-ai/grok-4.5[high]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 54,
    "solve_rate": 1.0,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 40413.40740740741,
    "tokens_per_success": 40413.40740740741,
    "conditional_tokens_among_solved_loops": 40413.40740740741,
    "mean_tokens_per_solved_task": 40413.40740740741,
    "mean_input_tokens_per_repair_loop": 36682.11111111111,
    "mean_output_tokens_per_repair_loop": 3731.296296296296,
    "mean_cache_read_tokens_per_repair_loop": 29660.444444444445,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 961.8888888888889,
    "mean_turns_per_repair_loop": 7.462962962962963,
    "p95_turns_per_repair_loop": 17.0,
    "mean_agent_steps_per_repair_loop": 7.462962962962963,
    "p95_agent_steps_per_repair_loop": 17.0,
    "mean_verifier_submissions_per_repair_loop": 1.0555555555555556,
    "mean_verifier_submissions_to_success": 1.0555555555555556,
    "stop_reasons": {
      "passed": 54
    },
    "total_model_spend_usd": 2.768112,
    "mean_cost_per_repair_loop": 0.05126133333333333,
    "p95_cost_per_repair_loop": 0.167036,
    "cpsc": 0.05126133333333333,
    "realized_cpsc": 0.05126133333333333,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 2.768112,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.05126133333333333,
    "mean_spend_per_solved_task": 0.05126133333333333,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.05112577777777778,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.00013555555555555543,
    "cost_delta_vs_gateway_ratio": 0.00265141307277042
  },
  {
    "model_config": "openrouter/z-ai/glm-5.2[high]",
    "total_trials": 54,
    "repair_loops": 54,
    "excluded_repair_loops": 0,
    "successes": 53,
    "solve_rate": 0.9814814814814815,
    "cap_hits": 0,
    "cap_hit_rate": 0.0,
    "mean_tokens_per_repair_loop": 52482.74074074074,
    "tokens_per_success": 53472.981132075474,
    "conditional_tokens_among_solved_loops": 52712.8679245283,
    "mean_tokens_per_solved_task": 52712.8679245283,
    "mean_input_tokens_per_repair_loop": 48078.27777777778,
    "mean_output_tokens_per_repair_loop": 4404.462962962963,
    "mean_cache_read_tokens_per_repair_loop": 39243.75925925926,
    "mean_cache_write_tokens_per_repair_loop": 0.0,
    "mean_reasoning_tokens_per_repair_loop": 0.0,
    "mean_turns_per_repair_loop": 11.037037037037036,
    "p95_turns_per_repair_loop": 19.0,
    "mean_agent_steps_per_repair_loop": 11.037037037037036,
    "p95_agent_steps_per_repair_loop": 19.0,
    "mean_verifier_submissions_per_repair_loop": 1.0925925925925926,
    "mean_verifier_submissions_to_success": 1.1132075471698113,
    "stop_reasons": {
      "agent_exit_repeatedformaterror": 1,
      "passed": 53
    },
    "total_model_spend_usd": 1.53864186,
    "mean_cost_per_repair_loop": 0.02849336777777778,
    "p95_cost_per_repair_loop": 0.08790534,
    "cpsc": 0.029030978490566038,
    "realized_cpsc": 0.029030978490566038,
    "reference_budget_cpsc": null,
    "escalation_cpsc": null,
    "total_reference_budget_charged_spend_usd": null,
    "total_realized_charged_spend_usd": 1.53864186,
    "total_escalation_charged_spend_usd": null,
    "reference_budget_complete_rows": 0,
    "replacement_cost_complete_rows": 0,
    "conditional_spend_among_solved_loops": 0.027443896981132077,
    "mean_spend_per_solved_task": 0.027443896981132077,
    "gateway_reported_repair_loops": 54,
    "mean_gateway_reported_cost_per_repair_loop": 0.022784761327407407,
    "mean_cost_delta_vs_gateway_per_repair_loop": 0.005708606450370372,
    "cost_delta_vs_gateway_ratio": 0.25054493081319157
  }
]
