{
  "schema_version": "decisions-1.1",
  "status": "preliminary",
  "suite_version": "1.0.0",
  "suite_hash": "b2d7e19cbe744ee0abddb95229d1fac9b9943f04dc5ed1ef826f641872fd0f8d",
  "protocol_hash": "d41bbc49c44ec0be17b1c5d079387440663bf8c81f5510ff7b92e7068a13489e",
  "campaign_hash": "aa23f0183149620613517fc86e83b5509474729de08c5917a6464dd089213b96",
  "recovery": null,
  "cases": 6,
  "families": 2,
  "items": 272,
  "categories": {
    "verdict": 18,
    "hypothesis_update": 36,
    "conclusion_check": 54,
    "evidence_relevance": 86,
    "next_step": 18,
    "model_routing": 60
  },
  "models": [
    {
      "model_id": "z-ai/glm-5.3-flash",
      "resolved_models": [
        "z-ai/glm-5.3-flash"
      ],
      "label": "GLM-5.3-Flash (FP8)",
      "backend": "openrouter_chat",
      "provider": "z-ai/fp8",
      "effort": "high",
      "repeats": 1,
      "cost_usd": 0.03443796,
      "cost_upper_bound_usd": 0.03443796,
      "mean_latency_seconds": 1.8470490547016893,
      "categories": {
        "verdict": {
          "accuracy": 50.0,
          "balanced_accuracy": 48.148148148148145,
          "constant_choice_baseline": 72.22222222222221,
          "attempts": 18,
          "invalid": 9,
          "critical_errors": 0,
          "brier": null,
          "probability_coverage": 0.0
        },
        "hypothesis_update": {
          "accuracy": 61.11111111111111,
          "balanced_accuracy": 68.51851851851852,
          "constant_choice_baseline": 72.22222222222221,
          "attempts": 36,
          "invalid": 10,
          "critical_errors": 0,
          "brier": null,
          "probability_coverage": 0.0
        },
        "conclusion_check": {
          "accuracy": 74.07407407407408,
          "balanced_accuracy": 76.38888888888889,
          "constant_choice_baseline": 66.66666666666666,
          "attempts": 54,
          "invalid": 13,
          "critical_errors": 1,
          "brier": null,
          "probability_coverage": 0.0
        },
        "evidence_relevance": {
          "accuracy": 80.23809523809524,
          "balanced_accuracy": 82.38636363636364,
          "constant_choice_baseline": 72.06349206349206,
          "attempts": 86,
          "invalid": 17,
          "critical_errors": 0,
          "brier": null,
          "probability_coverage": 0.0
        },
        "next_step": {
          "accuracy": 55.55555555555556,
          "balanced_accuracy": null,
          "constant_choice_baseline": 72.22222222222221,
          "attempts": 18,
          "invalid": 5,
          "critical_errors": 0,
          "brier": null,
          "probability_coverage": 0.0
        },
        "model_routing": {
          "accuracy": 96.66666666666667,
          "balanced_accuracy": 95.83333333333334,
          "constant_choice_baseline": 40.0,
          "attempts": 60,
          "invalid": 2,
          "critical_errors": 0,
          "brier": null,
          "probability_coverage": 0.0
        }
      }
    }
  ],
  "release_hash": "5a2f1e011a084fb47ed545e2f1e1b26f6ee469124b8393de8933d758a9464c1b"
}
