{
  "benchmark_version": "0.2.0",
  "generated_at": "2026-11-03T03:41:56.054176+03:00",
  "report_type": "paired_GPT_development_comparison",
  "headline_eligible": false,
  "condition": "A",
  "weight_view": "full",
  "suite_digest": "944ab7656cb18ccd7a44560ff9b2a9f75d48fcb3464a2049ec4dbde6ad30b8fc",
  "harness_digest": "6b0a481c0774120952a4f0ad5ce2cd29f4d695703d6f8f04f8570a40b9936ed2",
  "protocol_digest": "e49e01dd756ab9e02ea172fd532cfa717bd0eab96fa47c84d8c468c1f4203c7b",
  "manifest_digest": "605510de0ed1e019ebed041832364a193f02c6991482204da8baebd0e4f9f241",
  "reporter_digest": "c994a061b5a8c6eedf8927f775c10cc6c8e8cd717680093093948bfb689950a8",
  "status": "complete",
  "matrix_complete": true,
  "review_complete": true,
  "paired_inputs_match": true,
  "planned_sessions": 18,
  "launched_sessions": 18,
  "completed_scored_sessions": 18,
  "infrastructure_failures": 0,
  "inflight_sessions": 0,
  "missing_sessions": 0,
  "unknown_cost_sessions": 0,
  "source_sha256": "ebc788a2777e72b89b15ba94d021fb72c5db1a877bc570fb791745715b78ec46",
  "design_revision": "composed-circuit-v0.2.1",
  "scope": "Six preselected development cases per model, one attempt per case. No Repair trials.",
  "limitations": [
    "This small development comparison does not support a headline ranking or a combined capability score.",
    "Read is the mean of rule-recovery agreement and edit-effects before/after agreement, once per parent.",
    "Write accuracy alone is insufficient because the selected targets are class-imbalanced.",
    "Exact success uses the scored final cases. The two exact Write constructions also match all 1,024 domain entries.",
    "Median time includes the local session and API round trip. It is not isolated provider compute time.",
    "API costs are estimates for six scored sessions, using the run's recorded pricing and usage.",
    "The model names are provider aliases rather than immutable dated snapshots.",
    "Skill and family rows describe overlapping observations and must not be summed."
  ],
  "write_difficulty_control": {
    "name": "static_zero_start_no_edit",
    "source_actor": "metadata_prior",
    "recorded_condition": "B",
    "weight_view": "full",
    "same_access_model_score": false,
    "independent_tasks_added": 0,
    "selected_targets": 2,
    "exact_successes": 0,
    "exact_success": {
      "mean": 0.0,
      "n_sessions": 2,
      "n_parents": 2,
      "n_mechanisms": 2
    },
    "target_accuracy": {
      "mean": 0.701171875,
      "n_sessions": 2,
      "n_parents": 2,
      "n_mechanisms": 2
    },
    "affected_accuracy": {
      "mean": 0.0,
      "n_sessions": 2,
      "n_parents": 2,
      "n_mechanisms": 2
    },
    "retention_accuracy": {
      "mean": 1.0,
      "n_sessions": 2,
      "n_parents": 2,
      "n_mechanisms": 2
    },
    "retention_loss": {
      "mean": 0.0,
      "n_sessions": 2,
      "n_parents": 2,
      "n_mechanisms": 2
    },
    "evidence_digest": "e37c8b734e96ee244173f9d34ba9f0f3a2c805ca8d206c3de6bb992cffcccece",
    "interpretation": "Static difficulty control on the identical selected targets. Recorded in B, it reads no weights or feedback; it is neither an A model score nor additional independent tasks."
  },
  "models": [
    {
      "configuration": {
        "id": "openai-luna",
        "provider": "openai",
        "model": "openai/gpt-6-luna",
        "snapshot": "gpt-6-luna",
        "snapshot_kind": "provider-alias-no-dated-snapshot-published",
        "verified_at": "2026-11-03",
        "reasoning_effort": "high",
        "reasoning_summary": "none",
        "responses_api": true,
        "service_tier": "default",
        "pricing": {
          "input_per_million": 0.1,
          "cache_write_per_million": 0.125,
          "cache_read_per_million": 0.01,
          "output_per_million": 0.5,
          "source": "https://developers.openai.com/api/docs/models/gpt-6-luna",
          "verified_at": "2026-11-03",
          "cache_ttl": "30m_default"
        },
        "configuration_digest": "a83c85223298b6621af3350a1a3a2b24072ae6e7496308ae3be8ef923d1fcc20"
      },
      "audit_statuses": {
        "approved": 6
      },
      "scored_sessions": 6,
      "invalid_submissions": 0,
      "protocol_violations": 0,
      "tokens": {
        "input_tokens": 18,
        "input_tokens_cache_write": 310920,
        "input_tokens_cache_read": 0,
        "input_tokens_total": 310938,
        "output_tokens": 89132,
        "reasoning_tokens": 87169,
        "total_tokens": 400070
      },
      "estimated_api_cost_usd": 0.0834328,
      "session_wall_seconds": {
        "n": 6,
        "median": 117.29224322899972,
        "minimum": 56.99516683300317,
        "maximum": 247.399064792
      },
      "skills": [
        {
          "skill": "read",
          "scored_sessions": 4,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.44140625,
            "n_sessions": 4,
            "n_parents": 4,
            "n_mechanisms": 4
          },
          "metrics": {
            "affected_precision": {
              "mean": 0.5,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "affected_recall": {
              "mean": 0.5,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "after_agreement": {
              "mean": 0.4466145833333333,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "before_agreement": {
              "mean": 0.46484375,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "consequence_agreement": {
              "mean": 0.7056962025316456,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.411865234375,
              "n_sessions": 4,
              "n_parents": 4,
              "n_mechanisms": 4
            },
            "false_positive_rate": {
              "mean": 0.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "functional_agreement": {
              "mean": 0.4270833333333333,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "skill": "write",
          "scored_sessions": 2,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.71875,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_accuracy": {
              "mean": 0.07560975609756097,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_fraction": {
              "mean": 0.005433440355643368,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_norm": {
              "mean": 3.427827300200522,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.72265625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_change_rate": {
              "mean": 0.003552397868561279,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "parameters": {
              "mean": 4049.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_accuracy": {
              "mean": 0.9964476021314388,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_loss": {
              "mean": 0.0035523978685612634,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "target_accuracy": {
              "mean": 0.71875,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        }
      ],
      "families": [
        {
          "family": "compositional_synthesis",
          "scored_sessions": 2,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.71875,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_accuracy": {
              "mean": 0.07560975609756097,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_fraction": {
              "mean": 0.005433440355643368,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_norm": {
              "mean": 3.427827300200522,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.72265625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_change_rate": {
              "mean": 0.003552397868561279,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "parameters": {
              "mean": 4049.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_accuracy": {
              "mean": 0.9964476021314388,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_loss": {
              "mean": 0.0035523978685612634,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "target_accuracy": {
              "mean": 0.71875,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "family": "edit_effects",
          "scored_sessions": 2,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.45572916666666663,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_precision": {
              "mean": 0.5,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "affected_recall": {
              "mean": 0.5,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "after_agreement": {
              "mean": 0.4466145833333333,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "before_agreement": {
              "mean": 0.46484375,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "consequence_agreement": {
              "mean": 0.7056962025316456,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.39794921875,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_rate": {
              "mean": 0.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "family": "rule_recovery",
          "scored_sessions": 2,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.4270833333333333,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "exhaustive_agreement": {
              "mean": 0.42578125,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "functional_agreement": {
              "mean": 0.4270833333333333,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        }
      ]
    },
    {
      "configuration": {
        "id": "openai-terra",
        "provider": "openai",
        "model": "openai/gpt-6.1-terra",
        "snapshot": "gpt-6.1-terra",
        "snapshot_kind": "provider-alias-no-dated-snapshot-published",
        "verified_at": "2026-11-03",
        "reasoning_effort": "high",
        "reasoning_summary": "none",
        "responses_api": true,
        "service_tier": "default",
        "pricing": {
          "input_per_million": 1.0,
          "cache_write_per_million": 1.25,
          "cache_read_per_million": 0.05,
          "output_per_million": 5.0,
          "source": "https://developers.openai.com/api/docs/models/gpt-6.1-terra",
          "verified_at": "2026-11-03",
          "cache_ttl": "30m_default"
        },
        "configuration_digest": "ce3532abf1ad5bb61c94bee26237aa18c5204b6cead627928cf61a983c4ecb8e"
      },
      "audit_statuses": {
        "approved": 6
      },
      "scored_sessions": 6,
      "invalid_submissions": 0,
      "protocol_violations": 0,
      "tokens": {
        "input_tokens": 18,
        "input_tokens_cache_write": 310920,
        "input_tokens_cache_read": 0,
        "input_tokens_total": 310938,
        "output_tokens": 71501,
        "reasoning_tokens": 67367,
        "total_tokens": 382439
      },
      "estimated_api_cost_usd": 0.746173,
      "session_wall_seconds": {
        "n": 6,
        "median": 147.94193512450147,
        "minimum": 57.516623874995275,
        "maximum": 183.80786370800342
      },
      "skills": [
        {
          "skill": "read",
          "scored_sessions": 4,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.6595052083333333,
            "n_sessions": 4,
            "n_parents": 4,
            "n_mechanisms": 4
          },
          "metrics": {
            "affected_precision": {
              "mean": 0.6028645833333334,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "affected_recall": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "after_agreement": {
              "mean": 0.5813802083333333,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "before_agreement": {
              "mean": 0.486328125,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "consequence_agreement": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.5673828125,
              "n_sessions": 4,
              "n_parents": 4,
              "n_mechanisms": 4
            },
            "false_positive_rate": {
              "mean": 0.3971354166666667,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "functional_agreement": {
              "mean": 0.78515625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "skill": "write",
          "scored_sessions": 2,
          "exact_successes": 1,
          "functional_accuracy": {
            "mean": 0.97265625,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_accuracy": {
              "mean": 0.9609756097560975,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_fraction": {
              "mean": 0.023462583353914546,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_norm": {
              "mean": 11.456106690758634,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.97265625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_change_rate": {
              "mean": 0.023090586145648313,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "parameters": {
              "mean": 4049.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_accuracy": {
              "mean": 0.9769094138543517,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_loss": {
              "mean": 0.023090586145648295,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "target_accuracy": {
              "mean": 0.97265625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        }
      ],
      "families": [
        {
          "family": "compositional_synthesis",
          "scored_sessions": 2,
          "exact_successes": 1,
          "functional_accuracy": {
            "mean": 0.97265625,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_accuracy": {
              "mean": 0.9609756097560975,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_fraction": {
              "mean": 0.023462583353914546,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_norm": {
              "mean": 11.456106690758634,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.97265625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_change_rate": {
              "mean": 0.023090586145648313,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "parameters": {
              "mean": 4049.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_accuracy": {
              "mean": 0.9769094138543517,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_loss": {
              "mean": 0.023090586145648295,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "target_accuracy": {
              "mean": 0.97265625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "family": "edit_effects",
          "scored_sessions": 2,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.5338541666666666,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_precision": {
              "mean": 0.6028645833333334,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "affected_recall": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "after_agreement": {
              "mean": 0.5813802083333333,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "before_agreement": {
              "mean": 0.486328125,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "consequence_agreement": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.34375,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_rate": {
              "mean": 0.3971354166666667,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "family": "rule_recovery",
          "scored_sessions": 2,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.78515625,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "exhaustive_agreement": {
              "mean": 0.791015625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "functional_agreement": {
              "mean": 0.78515625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        }
      ]
    },
    {
      "configuration": {
        "id": "openai-sol",
        "provider": "openai",
        "model": "openai/gpt-6.1-sol",
        "snapshot": "gpt-6.1-sol",
        "snapshot_kind": "provider-alias-no-dated-snapshot-published",
        "verified_at": "2026-11-03",
        "reasoning_effort": "high",
        "reasoning_summary": "none",
        "responses_api": true,
        "service_tier": "default",
        "pricing": {
          "input_per_million": 2.0,
          "cache_write_per_million": 2.5,
          "cache_read_per_million": 0.1,
          "output_per_million": 10.0,
          "source": "https://developers.openai.com/api/docs/models/gpt-6.1-sol",
          "verified_at": "2026-11-03",
          "cache_ttl": "30m_default"
        },
        "configuration_digest": "cd4dc350f75e27f5933947ec0d100564f6ef6781c70e3f661a53493a20a4c758"
      },
      "audit_statuses": {
        "approved": 6
      },
      "scored_sessions": 6,
      "invalid_submissions": 0,
      "protocol_violations": 0,
      "tokens": {
        "input_tokens": 18,
        "input_tokens_cache_write": 310920,
        "input_tokens_cache_read": 0,
        "input_tokens_total": 310938,
        "output_tokens": 65122,
        "reasoning_tokens": 59867,
        "total_tokens": 376060
      },
      "estimated_api_cost_usd": 1.428556,
      "session_wall_seconds": {
        "n": 6,
        "median": 122.9375177289985,
        "minimum": 54.5847255839908,
        "maximum": 314.33287912500964
      },
      "skills": [
        {
          "skill": "read",
          "scored_sessions": 4,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.8766276041666667,
            "n_sessions": 4,
            "n_parents": 4,
            "n_mechanisms": 4
          },
          "metrics": {
            "affected_precision": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "affected_recall": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "after_agreement": {
              "mean": 0.8671875,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "before_agreement": {
              "mean": 0.8671875,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "consequence_agreement": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.875,
              "n_sessions": 4,
              "n_parents": 4,
              "n_mechanisms": 4
            },
            "false_positive_rate": {
              "mean": 0.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "functional_agreement": {
              "mean": 0.8860677083333334,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "skill": "write",
          "scored_sessions": 2,
          "exact_successes": 1,
          "functional_accuracy": {
            "mean": 0.9583333333333333,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_accuracy": {
              "mean": 0.9341463414634146,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_fraction": {
              "mean": 0.026179303531736232,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_norm": {
              "mean": 11.627715746420169,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.95703125,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_change_rate": {
              "mean": 0.03285968028419183,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "parameters": {
              "mean": 4049.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_accuracy": {
              "mean": 0.9671403197158082,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_loss": {
              "mean": 0.03285968028419184,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "target_accuracy": {
              "mean": 0.9583333333333333,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        }
      ],
      "families": [
        {
          "family": "compositional_synthesis",
          "scored_sessions": 2,
          "exact_successes": 1,
          "functional_accuracy": {
            "mean": 0.9583333333333333,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_accuracy": {
              "mean": 0.9341463414634146,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_fraction": {
              "mean": 0.026179303531736232,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "edit_norm": {
              "mean": 11.627715746420169,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.95703125,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_change_rate": {
              "mean": 0.03285968028419183,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "parameters": {
              "mean": 4049.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_accuracy": {
              "mean": 0.9671403197158082,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "retention_loss": {
              "mean": 0.03285968028419184,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "target_accuracy": {
              "mean": 0.9583333333333333,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "family": "edit_effects",
          "scored_sessions": 2,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.8671875,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "affected_precision": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "affected_recall": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "after_agreement": {
              "mean": 0.8671875,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "before_agreement": {
              "mean": 0.8671875,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "consequence_agreement": {
              "mean": 1.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "exhaustive_agreement": {
              "mean": 0.859375,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "false_positive_rate": {
              "mean": 0.0,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        },
        {
          "family": "rule_recovery",
          "scored_sessions": 2,
          "exact_successes": 0,
          "functional_accuracy": {
            "mean": 0.8860677083333334,
            "n_sessions": 2,
            "n_parents": 2,
            "n_mechanisms": 2
          },
          "metrics": {
            "exhaustive_agreement": {
              "mean": 0.890625,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            },
            "functional_agreement": {
              "mean": 0.8860677083333334,
              "n_sessions": 2,
              "n_parents": 2,
              "n_mechanisms": 2
            }
          }
        }
      ]
    }
  ],
  "historical_calibrations": [
    {
      "schema_version": "1",
      "report_type": "historical_foundation_calibration",
      "benchmark_version": "0.1.0",
      "generated_at": "2026-11-02T21:04:18.349475+00:00",
      "model": "openai/gpt-6-astra",
      "reasoning_effort": "high",
      "scored_sessions": 6,
      "reviewed_sessions": 6,
      "skills": {
        "read": {
          "tasks_passed": 2,
          "tasks_evaluated": 2,
          "attempts": 2,
          "task_pass_rate": 1.0
        },
        "repair": {
          "tasks_passed": 2,
          "tasks_evaluated": 2,
          "attempts": 2,
          "task_pass_rate": 1.0
        },
        "write": {
          "tasks_passed": 2,
          "tasks_evaluated": 2,
          "attempts": 2,
          "task_pass_rate": 1.0
        }
      },
      "estimated_api_cost_usd": 0.776872,
      "median_session_seconds": null,
      "headline_eligible": false,
      "freeze_eligible": false,
      "execution_harness_digests": [
        "56f9bb28f0d957ac5cd680a5c37d6ad2f942aed9ebe72d58c634e1bb1916b927",
        "b4e4fd26eb702683042c6abed37df907cbf8586b772c33a243fd39694603588b"
      ],
      "source_sha256": "be3615708c206d28ba75ed43b6a8799b050f12b909f714204abf97c26bba444e",
      "access": "Read without practice feedback. Repair used 24 and 16 practice examples; controller Write used 16 episodes; piecewise Write used none.",
      "interpretation": "Earlier v0.1 foundation tasks with different tool access. Task passes are not the current suite's agreement or exact-success metric. These rows cannot form a matched model comparison."
    }
  ]
}