{
  "date": "2026-09-14T01:18:07.274Z",
  "model": "gpt-6-astra",
  "effort": "low",
  "review_provenance": "Frozen twelve-case anchors previously inspected by the assistant against original PDF images. Not independent human ground truth. Narrative outputs reviewed by the assistant for separation of Q2 and H1 explanations.",
  "fact_rule": "Exact document/page from an eligible filing, metric, printed label, signed USD-million value, unit, accounting basis, start/end date and duration. Wrong period is incorrect even when amount matches. GAAP basis is an adapter assumption, not a blind visual result.",
  "calculation_tolerance": "1e-8 percentage points; unrelated to the reconciliation rounding allowance of 1.5 USD million.",
  "baseline": {
    "questions": 12,
    "passed": 11,
    "returned_reports": 11,
    "exact_facts": 10,
    "latency_samples": 12,
    "median_seconds": 80,
    "results": [
      {
        "id": "q2-revenue",
        "passed": false,
        "failures": [
          "409 session changed during parent-guarded runtime write"
        ],
        "exactFacts": 0,
        "seconds": 7,
        "tool_usage": []
      },
      {
        "id": "h1-revenue",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 87,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "q2-growth",
        "passed": true,
        "failures": [],
        "exactFacts": 2,
        "seconds": 117,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "net-income",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 68,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "h1-growth",
        "passed": true,
        "failures": [],
        "exactFacts": 2,
        "seconds": 95,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "common-income",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 73,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "gross-margin",
        "passed": true,
        "failures": [],
        "exactFacts": 2,
        "seconds": 88,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "annual-revenue",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 108,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "narrative",
        "passed": true,
        "failures": [],
        "exactFacts": 0,
        "seconds": 94,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "ambiguous",
        "passed": true,
        "failures": [],
        "exactFacts": 0,
        "seconds": 22,
        "status": "clarify",
        "tool_usage": []
      },
      {
        "id": "missing-company",
        "passed": true,
        "failures": [],
        "exactFacts": 0,
        "seconds": 18,
        "status": "unsupported",
        "tool_usage": []
      },
      {
        "id": "missing-period",
        "passed": true,
        "failures": [],
        "exactFacts": 0,
        "seconds": 21,
        "status": "unsupported",
        "tool_usage": []
      }
    ]
  },
  "enhanced": {
    "questions": 12,
    "passed": 12,
    "returned_reports": 12,
    "exact_facts": 11,
    "latency_samples": 11,
    "median_seconds": 67.653,
    "results": [
      {
        "id": "q2-revenue",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 72.288,
        "status": "answered",
        "tool_usage": [
          {
            "name": "find_table_cell",
            "outcome": "found",
            "duration_ms": 12,
            "reused": false
          }
        ]
      },
      {
        "id": "q2-growth",
        "passed": true,
        "failures": [],
        "exactFacts": 2,
        "seconds": 98.663,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "h1-revenue",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 67.653,
        "status": "answered",
        "tool_usage": [
          {
            "name": "find_table_cell",
            "outcome": "found",
            "duration_ms": 11,
            "reused": false
          }
        ]
      },
      {
        "id": "h1-growth",
        "passed": true,
        "failures": [],
        "exactFacts": 2,
        "seconds": 83.949,
        "status": "answered",
        "tool_usage": [
          {
            "name": "find_table_cell",
            "outcome": "found",
            "duration_ms": 15,
            "reused": false
          },
          {
            "name": "find_table_cell",
            "outcome": "found",
            "duration_ms": 10,
            "reused": false
          }
        ]
      },
      {
        "id": "net-income",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 76.286,
        "status": "answered",
        "tool_usage": [
          {
            "name": "find_table_cell",
            "outcome": "found",
            "duration_ms": 13,
            "reused": false
          }
        ]
      },
      {
        "id": "common-income",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 71.749,
        "status": "answered",
        "tool_usage": [
          {
            "name": "find_table_cell",
            "outcome": "found",
            "duration_ms": 11,
            "reused": false
          }
        ]
      },
      {
        "id": "gross-margin",
        "passed": true,
        "failures": [],
        "exactFacts": 2,
        "seconds": 66.526,
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "annual-revenue",
        "passed": true,
        "failures": [],
        "exactFacts": 1,
        "seconds": 66.117,
        "status": "answered",
        "tool_usage": [
          {
            "name": "find_table_cell",
            "outcome": "found",
            "duration_ms": 13,
            "reused": false
          }
        ]
      },
      {
        "id": "narrative",
        "passed": true,
        "failures": [],
        "exactFacts": 0,
        "seconds": null,
        "latency_note": "Local test process interrupted; completed original cloud turn recovered without resubmitting. Original start time unavailable.",
        "status": "answered",
        "tool_usage": []
      },
      {
        "id": "ambiguous",
        "passed": true,
        "failures": [],
        "exactFacts": 0,
        "seconds": 24.493,
        "status": "clarify",
        "tool_usage": []
      },
      {
        "id": "missing-company",
        "passed": true,
        "failures": [],
        "exactFacts": 0,
        "seconds": 32.214,
        "status": "unsupported",
        "tool_usage": []
      },
      {
        "id": "missing-period",
        "passed": true,
        "failures": [],
        "exactFacts": 0,
        "seconds": 21.138,
        "status": "unsupported",
        "tool_usage": []
      }
    ]
  }
}