{
  "snapshot": "2026-10-24",
  "provenance": "PAS7 normalized transcription of named primary publishers; no PAS7 inference tests. Missing values retained. Epoch data © Epoch AI, CC BY.",
  "models": {
    "gpt-6-astra": "GPT-6 Astra",
    "gpt-6-1-sol": "GPT-6.1 Sol",
    "claude-opus-5-5": "Claude Opus 5.5",
    "claude-sonnet-5-5": "Claude Sonnet 5.5"
  },
  "observations": [
    {
      "publisher": "Artificial Analysis",
      "metric": "Intelligence Index v4.3.2",
      "model": "gpt-6-astra",
      "value": 52.673669395513,
      "unit": "index",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Intelligence Index v4.3.2",
      "model": "gpt-6-1-sol",
      "value": 51.8332597011541,
      "unit": "index",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Intelligence Index v4.3.2",
      "model": "claude-opus-5-5",
      "value": 57.6223698102963,
      "unit": "index",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Intelligence Index v4.3.2",
      "model": "claude-sonnet-5-5",
      "value": 56.0001286580794,
      "unit": "index",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Briefcase v1.1",
      "model": "gpt-6-astra",
      "value": 1567.97,
      "unit": "Elo",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Briefcase v1.1",
      "model": "gpt-6-1-sol",
      "value": 1554.98,
      "unit": "Elo",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Briefcase v1.1",
      "model": "claude-opus-5-5",
      "value": 1801.02,
      "unit": "Elo",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Briefcase v1.1",
      "model": "claude-sonnet-5-5",
      "value": 1815.77,
      "unit": "Elo",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GDPval-AA 2.1",
      "model": "gpt-6-astra",
      "value": 1574.46,
      "unit": "Elo",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GDPval-AA 2.1",
      "model": "gpt-6-1-sol",
      "value": 1596.2,
      "unit": "Elo",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GDPval-AA 2.1",
      "model": "claude-opus-5-5",
      "value": 1866.27,
      "unit": "Elo",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GDPval-AA 2.1",
      "model": "claude-sonnet-5-5",
      "value": 1837.82,
      "unit": "Elo",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AutomationBench-AA partial",
      "model": "gpt-6-astra",
      "value": 68.49174806688876,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AutomationBench-AA partial",
      "model": "gpt-6-1-sol",
      "value": 64.86844882398171,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AutomationBench-AA partial",
      "model": "claude-opus-5-5",
      "value": 69.5386812863607,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AutomationBench-AA partial",
      "model": "claude-sonnet-5-5",
      "value": 71.76882128193995,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Terminal-Bench 4.0",
      "model": "gpt-6-astra",
      "value": 59.09090909090911,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Terminal-Bench 4.0",
      "model": "gpt-6-1-sol",
      "value": 56.0606060606061,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Terminal-Bench 4.0",
      "model": "claude-opus-5-5",
      "value": 59.5959595959596,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Terminal-Bench 4.0",
      "model": "claude-sonnet-5-5",
      "value": 63.636363636363605,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "FrontierCode",
      "model": "gpt-6-astra",
      "value": 46.9827666666667,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "FrontierCode",
      "model": "gpt-6-1-sol",
      "value": 49.1209333333333,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "FrontierCode",
      "model": "claude-opus-5-5",
      "value": 52.17309999999999,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "FrontierCode",
      "model": "claude-sonnet-5-5",
      "value": 45.8291333333333,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Terminal-Bench-Science",
      "model": "gpt-6-astra",
      "value": 63.3333333333333,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Terminal-Bench-Science",
      "model": "gpt-6-1-sol",
      "value": 58.0952380952381,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Terminal-Bench-Science",
      "model": "claude-opus-5-5",
      "value": 59.047619047619,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Terminal-Bench-Science",
      "model": "claude-sonnet-5-5",
      "value": 53.3333333333333,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "SciCode (under review)",
      "model": "gpt-6-astra",
      "value": 56.4814814814815,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "SciCode (under review)",
      "model": "gpt-6-1-sol",
      "value": 54.1666666666667,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "SciCode (under review)",
      "model": "claude-opus-5-5",
      "value": 66.8981481481482,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "SciCode (under review)",
      "model": "claude-sonnet-5-5",
      "value": 60.9953703703704,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Humanity’s Last Exam",
      "model": "gpt-6-astra",
      "value": 54.6802594995366,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Humanity’s Last Exam",
      "model": "gpt-6-1-sol",
      "value": 52.919369786839695,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Humanity’s Last Exam",
      "model": "claude-opus-5-5",
      "value": 61.353104726598694,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Humanity’s Last Exam",
      "model": "claude-sonnet-5-5",
      "value": 54.9582947173309,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "CritPt",
      "model": "gpt-6-astra",
      "value": 31.7142857142857,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "CritPt",
      "model": "gpt-6-1-sol",
      "value": 31.7142857142857,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "CritPt",
      "model": "claude-opus-5-5",
      "value": 31.7142857142857,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "CritPt",
      "model": "claude-sonnet-5-5",
      "value": 31.4285714285714,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GDPpdf all-pass",
      "model": "gpt-6-astra",
      "value": 31.0,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GDPpdf all-pass",
      "model": "gpt-6-1-sol",
      "value": 31.0,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GDPpdf all-pass",
      "model": "claude-opus-5-5",
      "value": 26.200000000000003,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GDPpdf all-pass",
      "model": "claude-sonnet-5-5",
      "value": 25.8,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Long Context Reasoning v1.1",
      "model": "gpt-6-astra",
      "value": 80.6666666666667,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Long Context Reasoning v1.1",
      "model": "gpt-6-1-sol",
      "value": 83.0,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Long Context Reasoning v1.1",
      "model": "claude-opus-5-5",
      "value": 84.6666666666667,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Long Context Reasoning v1.1",
      "model": "claude-sonnet-5-5",
      "value": 82.6666666666667,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Multilingual Long Context Reasoning",
      "model": "gpt-6-astra",
      "value": 41.6666666666667,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Multilingual Long Context Reasoning",
      "model": "gpt-6-1-sol",
      "value": 38.3333333333333,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Multilingual Long Context Reasoning",
      "model": "claude-opus-5-5",
      "value": 27.7777777777778,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Multilingual Long Context Reasoning",
      "model": "claude-sonnet-5-5",
      "value": 41.6666666666667,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Analyst Agent",
      "model": "gpt-6-astra",
      "value": 51.24999999999999,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Analyst Agent",
      "model": "gpt-6-1-sol",
      "value": 50.0,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Analyst Agent",
      "model": "claude-opus-5-5",
      "value": 56.25,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Analyst Agent",
      "model": "claude-sonnet-5-5",
      "value": 57.49999999999999,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "MMMU-Pro",
      "model": "gpt-6-astra",
      "value": 86.878612716763,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "MMMU-Pro",
      "model": "gpt-6-1-sol",
      "value": 85.9537572254335,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "MMMU-Pro",
      "model": "claude-opus-5-5",
      "value": 87.6878612716763,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "MMMU-Pro",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Omniscience",
      "model": "gpt-6-astra",
      "value": 43.4,
      "unit": "index",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Omniscience",
      "model": "gpt-6-1-sol",
      "value": 41.53333333333333,
      "unit": "index",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Omniscience",
      "model": "claude-opus-5-5",
      "value": 46.416666666666664,
      "unit": "index",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Omniscience",
      "model": "claude-sonnet-5-5",
      "value": 32.3,
      "unit": "index",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Omniscience accuracy",
      "model": "gpt-6-astra",
      "value": 62.6,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Omniscience accuracy",
      "model": "gpt-6-1-sol",
      "value": 62.1,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Omniscience accuracy",
      "model": "claude-opus-5-5",
      "value": 66.21666666666667,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA Omniscience accuracy",
      "model": "claude-sonnet-5-5",
      "value": 53.949999999999996,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA 3D",
      "model": "gpt-6-astra",
      "value": 1164.1313,
      "unit": "Elo",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA 3D",
      "model": "gpt-6-1-sol",
      "value": 1119.3508,
      "unit": "Elo",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA 3D",
      "model": "claude-opus-5-5",
      "value": 1196.3999,
      "unit": "Elo",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AA 3D",
      "model": "claude-sonnet-5-5",
      "value": 1198.8306,
      "unit": "Elo",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GPQA Diamond (legacy)",
      "model": "gpt-6-astra",
      "value": 96.0606060606061,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GPQA Diamond (legacy)",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GPQA Diamond (legacy)",
      "model": "claude-opus-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "GPQA Diamond (legacy)",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AIME 2025",
      "model": "gpt-6-astra",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AIME 2025",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AIME 2025",
      "model": "claude-opus-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "AIME 2025",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "MMLU-Pro",
      "model": "gpt-6-astra",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "MMLU-Pro",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "MMLU-Pro",
      "model": "claude-opus-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "MMLU-Pro",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "LiveCodeBench",
      "model": "gpt-6-astra",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "LiveCodeBench",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "LiveCodeBench",
      "model": "claude-opus-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "LiveCodeBench",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "ITBench SRE",
      "model": "gpt-6-astra",
      "value": 48.5875706214689,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "ITBench SRE",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "ITBench SRE",
      "model": "claude-opus-5-5",
      "value": 38.229755178907695,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "ITBench SRE",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Published AA evaluation; null means no score in the checked record. SciCode is marked under review.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Weighted Intelligence Index cost per task",
      "model": "gpt-6-astra",
      "value": 3.2575003134834164,
      "unit": "USD/task",
      "configuration": "max",
      "interval": null,
      "notes": "Specific weighted workload, including reasoning and cache costs; not a universal request price.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Weighted Intelligence Index cost per task",
      "model": "gpt-6-1-sol",
      "value": 0.7241670655535033,
      "unit": "USD/task",
      "configuration": "max",
      "interval": null,
      "notes": "Specific weighted workload, including reasoning and cache costs; not a universal request price.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Weighted Intelligence Index cost per task",
      "model": "claude-opus-5-5",
      "value": 5.982012019521066,
      "unit": "USD/task",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Specific weighted workload, including reasoning and cache costs; not a universal request price.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Weighted Intelligence Index cost per task",
      "model": "claude-sonnet-5-5",
      "value": 5.4608525948098325,
      "unit": "USD/task",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Specific weighted workload, including reasoning and cache costs; not a universal request price.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Output throughput",
      "model": "gpt-6-astra",
      "value": 58.1181143902355,
      "unit": "tokens/second",
      "configuration": "max",
      "interval": null,
      "notes": "Output throughput is not end-to-end task latency.",
      "source": "https://artificialanalysis.ai/models/gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Output throughput",
      "model": "gpt-6-1-sol",
      "value": 57.3369712661238,
      "unit": "tokens/second",
      "configuration": "max",
      "interval": null,
      "notes": "Output throughput is not end-to-end task latency.",
      "source": "https://artificialanalysis.ai/models/gpt-6-1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Output throughput",
      "model": "claude-opus-5-5",
      "value": 96.2932485108269,
      "unit": "tokens/second",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Output throughput is not end-to-end task latency.",
      "source": "https://artificialanalysis.ai/models/claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Artificial Analysis",
      "metric": "Output throughput",
      "model": "claude-sonnet-5-5",
      "value": 132.7139358127,
      "unit": "tokens/second",
      "configuration": "max; default fallback",
      "interval": null,
      "notes": "Output throughput is not end-to-end task latency.",
      "source": "https://artificialanalysis.ai/models/claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Text Arena",
      "model": "gpt-6-astra",
      "value": 1476,
      "unit": "rating",
      "configuration": "max",
      "interval": "±6",
      "notes": "Human preference; board snapshot 2026-10-22. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/chat/text",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Text Arena",
      "model": "gpt-6-1-sol",
      "value": 1480,
      "unit": "rating",
      "configuration": "max",
      "interval": "±7",
      "notes": "Human preference; board snapshot 2026-10-22. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/chat/text",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Text Arena",
      "model": "claude-opus-5-5",
      "value": 1509,
      "unit": "rating",
      "configuration": "high",
      "interval": "±7",
      "notes": "Human preference; board snapshot 2026-10-22. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/chat/text",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Text Arena",
      "model": "claude-sonnet-5-5",
      "value": 1478,
      "unit": "rating",
      "configuration": "xhigh",
      "interval": "±7",
      "notes": "Human preference; board snapshot 2026-10-22. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/chat/text",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Vision Arena",
      "model": "gpt-6-astra",
      "value": 1285,
      "unit": "rating",
      "configuration": "max",
      "interval": "±10",
      "notes": "Human preference; board snapshot 2026-10-19. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/chat/vision",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Vision Arena",
      "model": "gpt-6-1-sol",
      "value": 1294,
      "unit": "rating",
      "configuration": "max",
      "interval": "±13",
      "notes": "Human preference; board snapshot 2026-10-19. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/chat/vision",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Vision Arena",
      "model": "claude-opus-5-5",
      "value": 1293,
      "unit": "rating",
      "configuration": "high",
      "interval": "±12",
      "notes": "Human preference; board snapshot 2026-10-19. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/chat/vision",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Vision Arena",
      "model": "claude-sonnet-5-5",
      "value": 1265,
      "unit": "rating",
      "configuration": "xhigh",
      "interval": "±13",
      "notes": "Human preference; board snapshot 2026-10-19. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/chat/vision",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "WebDev Arena",
      "model": "gpt-6-astra",
      "value": 1774,
      "unit": "rating",
      "configuration": "max",
      "interval": "±10",
      "notes": "Human preference; board snapshot 2026-10-22. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/code/webdev",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "WebDev Arena",
      "model": "gpt-6-1-sol",
      "value": 1749,
      "unit": "rating",
      "configuration": "max",
      "interval": "±12",
      "notes": "Human preference; board snapshot 2026-10-22. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/code/webdev",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "WebDev Arena",
      "model": "claude-opus-5-5",
      "value": 1807,
      "unit": "rating",
      "configuration": "max",
      "interval": "±13",
      "notes": "Human preference; board snapshot 2026-10-22. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/code/webdev",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "WebDev Arena",
      "model": "claude-sonnet-5-5",
      "value": 1769,
      "unit": "rating",
      "configuration": "xhigh",
      "interval": "±12",
      "notes": "Human preference; board snapshot 2026-10-22. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/code/webdev",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Image-to-WebDev Arena",
      "model": "gpt-6-astra",
      "value": 1726,
      "unit": "rating",
      "configuration": "max",
      "interval": "±12",
      "notes": "Human preference; board snapshot retrieved 2026-10-24. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/code/image-to-webdev",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Image-to-WebDev Arena",
      "model": "gpt-6-1-sol",
      "value": 1702,
      "unit": "rating",
      "configuration": "max",
      "interval": "±15",
      "notes": "Human preference; board snapshot retrieved 2026-10-24. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/code/image-to-webdev",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Image-to-WebDev Arena",
      "model": "claude-opus-5-5",
      "value": 1753,
      "unit": "rating",
      "configuration": "max",
      "interval": "±24",
      "notes": "Human preference; board snapshot retrieved 2026-10-24. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/code/image-to-webdev",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Arena",
      "metric": "Image-to-WebDev Arena",
      "model": "claude-sonnet-5-5",
      "value": 1737,
      "unit": "rating",
      "configuration": "xhigh",
      "interval": "±17",
      "notes": "Human preference; board snapshot retrieved 2026-10-24. Different efforts; overlapping uncertainty is not a decisive win.",
      "source": "https://arena.ai/leaderboard/code/image-to-webdev",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "BioMysteryBench",
      "model": "gpt-6-astra",
      "value": 79.26,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.98",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | BioMysteryBench: 57 fallbacks (21.11%). 6 provider refusals (2.22%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "BioMysteryBench",
      "model": "gpt-6-1-sol",
      "value": 79.63,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.74",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | BioMysteryBench: 57 fallbacks (21.11%). 6 provider refusals (2.22%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "BioMysteryBench",
      "model": "claude-opus-5-5",
      "value": 79.26,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.98",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | BioMysteryBench: 57 fallbacks (21.11%). 6 provider refusals (2.22%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "BioMysteryBench",
      "model": "claude-sonnet-5-5",
      "value": 81.11,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.64",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | BioMysteryBench: 57 fallbacks (21.11%). 6 provider refusals (2.22%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "CUA-bench",
      "model": "gpt-6-astra",
      "value": 19.17,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "CUA-bench",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "CUA-bench",
      "model": "claude-opus-5-5",
      "value": 14.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "CUA-bench",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ClaimBench",
      "model": "gpt-6-astra",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ClaimBench",
      "model": "gpt-6-1-sol",
      "value": 100.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ClaimBench",
      "model": "claude-opus-5-5",
      "value": 100.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ClaimBench",
      "model": "claude-sonnet-5-5",
      "value": 100.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Code Migration v1.1",
      "model": "gpt-6-astra",
      "value": 80.58,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.34",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Code Migration v1.1: 42 fallbacks (32.31%). 4 provider refusals (3.08%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Code Migration v1.1",
      "model": "gpt-6-1-sol",
      "value": 77.32,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.98",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Code Migration v1.1: 42 fallbacks (32.31%). 4 provider refusals (3.08%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Code Migration v1.1",
      "model": "claude-opus-5-5",
      "value": 81.01,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.21",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Code Migration v1.1: 42 fallbacks (32.31%). 4 provider refusals (3.08%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Code Migration v1.1",
      "model": "claude-sonnet-5-5",
      "value": 86.88,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.52",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Code Migration v1.1: 42 fallbacks (32.31%). 4 provider refusals (3.08%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "CyberBench v1.1",
      "model": "gpt-6-astra",
      "value": 41.07,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.56",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | CyberBench v1.1: 61 fallbacks (52.59%). | CyberBench v1.1: 54 fallbacks (46.55%). 8 provider refusals (6.90%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "CyberBench v1.1",
      "model": "gpt-6-1-sol",
      "value": 39.29,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.74",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | CyberBench v1.1: 61 fallbacks (52.59%). | CyberBench v1.1: 54 fallbacks (46.55%). 8 provider refusals (6.90%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "CyberBench v1.1",
      "model": "claude-opus-5-5",
      "value": 55.36,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±5.13",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | CyberBench v1.1: 61 fallbacks (52.59%). | CyberBench v1.1: 54 fallbacks (46.55%). 8 provider refusals (6.90%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "CyberBench v1.1",
      "model": "claude-sonnet-5-5",
      "value": 59.58,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±5.21",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | CyberBench v1.1: 61 fallbacks (52.59%). | CyberBench v1.1: 54 fallbacks (46.55%). 8 provider refusals (6.90%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "EMB v1.1",
      "model": "gpt-6-astra",
      "value": 72.68,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.30",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "EMB v1.1",
      "model": "gpt-6-1-sol",
      "value": 71.09,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.75",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "EMB v1.1",
      "model": "claude-opus-5-5",
      "value": 75.84,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.41",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "EMB v1.1",
      "model": "claude-sonnet-5-5",
      "value": 75.9,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.47",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Finance Agent v2.1",
      "model": "gpt-6-astra",
      "value": 60.43,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.12",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Finance Agent v2.1: 5 fallbacks (1.11%). | Finance Agent v2.1: 32 fallbacks (7.13%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Finance Agent v2.1",
      "model": "gpt-6-1-sol",
      "value": 61.82,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.11",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Finance Agent v2.1: 5 fallbacks (1.11%). | Finance Agent v2.1: 32 fallbacks (7.13%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Finance Agent v2.1",
      "model": "claude-opus-5-5",
      "value": 67.85,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.05",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Finance Agent v2.1: 5 fallbacks (1.11%). | Finance Agent v2.1: 32 fallbacks (7.13%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Finance Agent v2.1",
      "model": "claude-sonnet-5-5",
      "value": 67.28,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.02",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Finance Agent v2.1: 5 fallbacks (1.11%). | Finance Agent v2.1: 32 fallbacks (7.13%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "HLAB-Vals-1.1",
      "model": "gpt-6-astra",
      "value": 7.08,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.90",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "HLAB-Vals-1.1",
      "model": "gpt-6-1-sol",
      "value": 6.25,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.83",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "HLAB-Vals-1.1",
      "model": "claude-opus-5-5",
      "value": 9.17,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.37",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "HLAB-Vals-1.1",
      "model": "claude-sonnet-5-5",
      "value": 10.42,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.57",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "IOI",
      "model": "gpt-6-astra",
      "value": 100.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | IOI: 1 fallbacks (5.56%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "IOI",
      "model": "gpt-6-1-sol",
      "value": 96.89,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.11",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | IOI: 1 fallbacks (5.56%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "IOI",
      "model": "claude-opus-5-5",
      "value": 95.06,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±4.94",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | IOI: 1 fallbacks (5.56%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "IOI",
      "model": "claude-sonnet-5-5",
      "value": 83.06,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.74",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | IOI: 1 fallbacks (5.56%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Legal Research Bench",
      "model": "gpt-6-astra",
      "value": 39.42,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.40",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Legal Research Bench: 5 fallbacks (2.40%). 11 provider refusals (5.29%), scored as failures. | Legal Research Bench: 11 fallbacks (5.29%). 3 provider refusals (1.44%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Legal Research Bench",
      "model": "gpt-6-1-sol",
      "value": 38.46,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.38",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Legal Research Bench: 5 fallbacks (2.40%). 11 provider refusals (5.29%), scored as failures. | Legal Research Bench: 11 fallbacks (5.29%). 3 provider refusals (1.44%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Legal Research Bench",
      "model": "claude-opus-5-5",
      "value": 50.48,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.48",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Legal Research Bench: 5 fallbacks (2.40%). 11 provider refusals (5.29%), scored as failures. | Legal Research Bench: 11 fallbacks (5.29%). 3 provider refusals (1.44%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Legal Research Bench",
      "model": "claude-sonnet-5-5",
      "value": 48.08,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.47",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Legal Research Bench: 5 fallbacks (2.40%). 11 provider refusals (5.29%), scored as failures. | Legal Research Bench: 11 fallbacks (5.29%). 3 provider refusals (1.44%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MedCode",
      "model": "gpt-6-astra",
      "value": 48.49,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.13",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MedCode",
      "model": "gpt-6-1-sol",
      "value": 48.84,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.12",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MedCode",
      "model": "claude-opus-5-5",
      "value": 49.8,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.27",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MedCode",
      "model": "claude-sonnet-5-5",
      "value": 52.92,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.12",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MedScribe",
      "model": "gpt-6-astra",
      "value": 87.91,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.94",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MedScribe",
      "model": "gpt-6-1-sol",
      "value": 86.45,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.90",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MedScribe",
      "model": "claude-opus-5-5",
      "value": 91.43,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.93",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MedScribe",
      "model": "claude-sonnet-5-5",
      "value": 91.1,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.96",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MysteryMechanism",
      "model": "gpt-6-astra",
      "value": 53.15,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.36",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | MysteryMechanism: 1 fallbacks (0.45%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MysteryMechanism",
      "model": "gpt-6-1-sol",
      "value": 46.4,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.35",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | MysteryMechanism: 1 fallbacks (0.45%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MysteryMechanism",
      "model": "claude-opus-5-5",
      "value": 49.55,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.36",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | MysteryMechanism: 1 fallbacks (0.45%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "MysteryMechanism",
      "model": "claude-sonnet-5-5",
      "value": 49.1,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.36",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | MysteryMechanism: 1 fallbacks (0.45%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ProgramBench",
      "model": "gpt-6-astra",
      "value": 5.5,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.62",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | ProgramBench: 24 fallbacks (12.00%). 1 provider refusals (0.50%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ProgramBench",
      "model": "gpt-6-1-sol",
      "value": 3.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.21",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | ProgramBench: 24 fallbacks (12.00%). 1 provider refusals (0.50%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ProgramBench",
      "model": "claude-opus-5-5",
      "value": 18.5,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.75",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | ProgramBench: 24 fallbacks (12.00%). 1 provider refusals (0.50%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ProgramBench",
      "model": "claude-sonnet-5-5",
      "value": 6.5,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.75",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | ProgramBench: 24 fallbacks (12.00%). 1 provider refusals (0.50%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ProofBench v1.1",
      "model": "gpt-6-astra",
      "value": 99.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ProofBench v1.1",
      "model": "gpt-6-1-sol",
      "value": 99.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ProofBench v1.1",
      "model": "claude-opus-5-5",
      "value": 100.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "ProofBench v1.1",
      "model": "claude-sonnet-5-5",
      "value": 100.0,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Public Benefits Bench v1.1",
      "model": "gpt-6-astra",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Public Benefits Bench v1.1: 3 fallbacks (1.30%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Public Benefits Bench v1.1",
      "model": "gpt-6-1-sol",
      "value": 59.34,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.28",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Public Benefits Bench v1.1: 3 fallbacks (1.30%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Public Benefits Bench v1.1",
      "model": "claude-opus-5-5",
      "value": 70.64,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.19",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Public Benefits Bench v1.1: 3 fallbacks (1.30%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Public Benefits Bench v1.1",
      "model": "claude-sonnet-5-5",
      "value": 67.19,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.22",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Public Benefits Bench v1.1: 3 fallbacks (1.30%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Robo-bench",
      "model": "gpt-6-astra",
      "value": 23.33,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Robo-bench",
      "model": "gpt-6-1-sol",
      "value": 28.33,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Robo-bench",
      "model": "claude-opus-5-5",
      "value": 14.58,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Robo-bench",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SAGE",
      "model": "gpt-6-astra",
      "value": 46.37,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.44",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SAGE",
      "model": "gpt-6-1-sol",
      "value": 46.54,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.46",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SAGE",
      "model": "claude-opus-5-5",
      "value": 45.83,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.36",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SAGE",
      "model": "claude-sonnet-5-5",
      "value": 51.77,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.41",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SRE Bench",
      "model": "gpt-6-astra",
      "value": 56.87,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.07",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | SRE Bench: 217 fallbacks (82.82%). 27 provider refusals (10.30%), scored as failures. | SRE Bench: 126 fallbacks (48.09%). 41 provider refusals (15.65%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SRE Bench",
      "model": "gpt-6-1-sol",
      "value": 50.76,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.10",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | SRE Bench: 217 fallbacks (82.82%). 27 provider refusals (10.30%), scored as failures. | SRE Bench: 126 fallbacks (48.09%). 41 provider refusals (15.65%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SRE Bench",
      "model": "claude-opus-5-5",
      "value": 33.59,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.92",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | SRE Bench: 217 fallbacks (82.82%). 27 provider refusals (10.30%), scored as failures. | SRE Bench: 126 fallbacks (48.09%). 41 provider refusals (15.65%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SRE Bench",
      "model": "claude-sonnet-5-5",
      "value": 30.15,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.84",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | SRE Bench: 217 fallbacks (82.82%). 27 provider refusals (10.30%), scored as failures. | SRE Bench: 126 fallbacks (48.09%). 41 provider refusals (15.65%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SWE-bench Multimodal",
      "model": "gpt-6-astra",
      "value": 43.33,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.26",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SWE-bench Multimodal",
      "model": "gpt-6-1-sol",
      "value": 43.54,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.27",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SWE-bench Multimodal",
      "model": "claude-opus-5-5",
      "value": 70.21,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.09",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "SWE-bench Multimodal",
      "model": "claude-sonnet-5-5",
      "value": 58.33,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.25",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Tax Agent Bench",
      "model": "gpt-6-astra",
      "value": 20.72,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.92",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Tax Agent Bench: 1 fallbacks (0.52%). 1 provider refusals (0.52%), scored as failures. | Tax Agent Bench: 8 fallbacks (4.15%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Tax Agent Bench",
      "model": "gpt-6-1-sol",
      "value": 21.75,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.98",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Tax Agent Bench: 1 fallbacks (0.52%). 1 provider refusals (0.52%), scored as failures. | Tax Agent Bench: 8 fallbacks (4.15%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Tax Agent Bench",
      "model": "claude-opus-5-5",
      "value": 45.57,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.59",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Tax Agent Bench: 1 fallbacks (0.52%). 1 provider refusals (0.52%), scored as failures. | Tax Agent Bench: 8 fallbacks (4.15%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Tax Agent Bench",
      "model": "claude-sonnet-5-5",
      "value": 41.96,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±3.56",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Tax Agent Bench: 1 fallbacks (0.52%). 1 provider refusals (0.52%), scored as failures. | Tax Agent Bench: 8 fallbacks (4.15%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Terminal-Bench 4.0",
      "model": "gpt-6-astra",
      "value": 59.6,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±4.40",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Terminal-Bench 4.0: 22 fallbacks (11.11%). | Terminal-Bench 4.0: 4 fallbacks (2.02%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Terminal-Bench 4.0",
      "model": "gpt-6-1-sol",
      "value": 55.05,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.82",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Terminal-Bench 4.0: 22 fallbacks (11.11%). | Terminal-Bench 4.0: 4 fallbacks (2.02%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Terminal-Bench 4.0",
      "model": "claude-opus-5-5",
      "value": 65.15,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Terminal-Bench 4.0: 22 fallbacks (11.11%). | Terminal-Bench 4.0: 4 fallbacks (2.02%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Terminal-Bench 4.0",
      "model": "claude-sonnet-5-5",
      "value": 64.14,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.01",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Terminal-Bench 4.0: 22 fallbacks (11.11%). | Terminal-Bench 4.0: 4 fallbacks (2.02%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Terminal-Bench-Science",
      "model": "gpt-6-astra",
      "value": 62.86,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±5.82",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Terminal-Bench-Science: 2 fallbacks (2.86%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Terminal-Bench-Science",
      "model": "gpt-6-1-sol",
      "value": 52.86,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±6.01",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Terminal-Bench-Science: 2 fallbacks (2.86%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Terminal-Bench-Science",
      "model": "claude-opus-5-5",
      "value": 47.14,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±6.01",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Terminal-Bench-Science: 2 fallbacks (2.86%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Terminal-Bench-Science",
      "model": "claude-sonnet-5-5",
      "value": 47.14,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±6.01",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Terminal-Bench-Science: 2 fallbacks (2.86%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Time Horizon Index: KSP",
      "model": "gpt-6-astra",
      "value": 90.5,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Time Horizon Index: KSP",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Time Horizon Index: KSP",
      "model": "claude-opus-5-5",
      "value": 91.33,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Time Horizon Index: KSP",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Finance Index",
      "model": "gpt-6-astra",
      "value": 55.64,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.38",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Finance Index: 6 fallbacks (0.81%). 1 provider refusals (0.13%), scored as failures. | Vals Finance Index: 40 fallbacks (5.38%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Finance Index",
      "model": "gpt-6-1-sol",
      "value": 55.81,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.50",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Finance Index: 6 fallbacks (0.81%). 1 provider refusals (0.13%), scored as failures. | Vals Finance Index: 40 fallbacks (5.38%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Finance Index",
      "model": "claude-opus-5-5",
      "value": 65.59,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.48",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Finance Index: 6 fallbacks (0.81%). 1 provider refusals (0.13%), scored as failures. | Vals Finance Index: 40 fallbacks (5.38%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Finance Index",
      "model": "claude-sonnet-5-5",
      "value": 64.53,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.48",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Finance Index: 6 fallbacks (0.81%). 1 provider refusals (0.13%), scored as failures. | Vals Finance Index: 40 fallbacks (5.38%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Index",
      "model": "gpt-6-astra",
      "value": 65.02,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.10",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Index: 59 fallbacks (4.28%). 14 provider refusals (1.01%), scored as failures. | Vals Index: 55 fallbacks (3.99%). 3 provider refusals (0.22%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Index",
      "model": "gpt-6-1-sol",
      "value": 63.94,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.07",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Index: 59 fallbacks (4.28%). 14 provider refusals (1.01%), scored as failures. | Vals Index: 55 fallbacks (3.99%). 3 provider refusals (0.22%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Index",
      "model": "claude-opus-5-5",
      "value": 70.18,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.00",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Index: 59 fallbacks (4.28%). 14 provider refusals (1.01%), scored as failures. | Vals Index: 55 fallbacks (3.99%). 3 provider refusals (0.22%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Index",
      "model": "claude-sonnet-5-5",
      "value": 70.85,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±0.93",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Index: 59 fallbacks (4.28%). 14 provider refusals (1.01%), scored as failures. | Vals Index: 55 fallbacks (3.99%). 3 provider refusals (0.22%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Legal Index",
      "model": "gpt-6-astra",
      "value": 68.39,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.74",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Legal Index: 6 fallbacks (1.15%). 12 provider refusals (2.30%), scored as failures. | Vals Legal Index: 19 fallbacks (3.65%). 3 provider refusals (0.58%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Legal Index",
      "model": "gpt-6-1-sol",
      "value": 68.72,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.77",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Legal Index: 6 fallbacks (1.15%). 12 provider refusals (2.30%), scored as failures. | Vals Legal Index: 19 fallbacks (3.65%). 3 provider refusals (0.58%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Legal Index",
      "model": "claude-opus-5-5",
      "value": 72.33,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.84",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Legal Index: 6 fallbacks (1.15%). 12 provider refusals (2.30%), scored as failures. | Vals Legal Index: 19 fallbacks (3.65%). 3 provider refusals (0.58%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Legal Index",
      "model": "claude-sonnet-5-5",
      "value": 68.02,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.81",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Legal Index: 6 fallbacks (1.15%). 12 provider refusals (2.30%), scored as failures. | Vals Legal Index: 19 fallbacks (3.65%). 3 provider refusals (0.58%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals RSI Index",
      "model": "gpt-6-astra",
      "value": 27.07,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals RSI Index",
      "model": "gpt-6-1-sol",
      "value": 26.62,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals RSI Index",
      "model": "claude-opus-5-5",
      "value": 37.31,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals RSI Index",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Science Index",
      "model": "gpt-6-astra",
      "value": 65.09,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.26",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Science Index: 60 fallbacks (10.68%). 6 provider refusals (1.07%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Science Index",
      "model": "gpt-6-1-sol",
      "value": 59.63,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.31",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Science Index: 60 fallbacks (10.68%). 6 provider refusals (1.07%), scored as failures.",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Science Index",
      "model": "claude-opus-5-5",
      "value": 58.65,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.32",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Science Index: 60 fallbacks (10.68%). 6 provider refusals (1.07%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vals Science Index",
      "model": "claude-sonnet-5-5",
      "value": 59.12,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.31",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vals Science Index: 60 fallbacks (10.68%). 6 provider refusals (1.07%), scored as failures.",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vibe Code Bench 1-100",
      "model": "gpt-6-astra",
      "value": 27.64,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±4.08",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vibe Code Bench 1-100: 8 fallbacks (16.00%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vibe Code Bench 1-100",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vibe Code Bench 1-100: 8 fallbacks (16.00%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vibe Code Bench 1-100",
      "model": "claude-opus-5-5",
      "value": 30.36,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±4.83",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vibe Code Bench 1-100: 8 fallbacks (16.00%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vibe Code Bench 1-100",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": null,
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vibe Code Bench 1-100: 8 fallbacks (16.00%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vibe Code Bench v1.1",
      "model": "gpt-6-astra",
      "value": 89.59,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±2.17",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vibe Code Bench v1.1: 4 fallbacks (8.00%).",
      "source": "https://www.vals.ai/models/openai_gpt-6-astra",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vibe Code Bench v1.1",
      "model": "gpt-6-1-sol",
      "value": 88.93,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.89",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vibe Code Bench v1.1: 4 fallbacks (8.00%).",
      "source": "https://www.vals.ai/models/openai_gpt-6.1-sol",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vibe Code Bench v1.1",
      "model": "claude-opus-5-5",
      "value": 90.29,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.53",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vibe Code Bench v1.1: 4 fallbacks (8.00%).",
      "source": "https://www.vals.ai/models/anthropic_claude-opus-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Vals AI",
      "metric": "Vibe Code Bench v1.1",
      "model": "claude-sonnet-5-5",
      "value": 92.39,
      "unit": "percent",
      "configuration": "max (default; benchmark overrides possible)",
      "interval": "±1.26",
      "notes": "Server-side fallbacks included; consult publisher per-task breakdown. | Vibe Code Bench v1.1: 4 fallbacks (8.00%).",
      "source": "https://www.vals.ai/models/anthropic_claude-sonnet-5-5",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "FrontierMath tiers 1–3 v2",
      "model": "gpt-6-astra",
      "value": 93.6842105263158,
      "unit": "percent",
      "configuration": "max",
      "interval": "SE 1.4434038295343787",
      "notes": "Epoch AI internal runs; mean_score; uncertainty is standard error, not a confidence interval.",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "FrontierMath tiers 1–3 v2",
      "model": "gpt-6-1-sol",
      "value": 93.6842105263158,
      "unit": "percent",
      "configuration": "max",
      "interval": "SE 1.4434038295343787",
      "notes": "Epoch AI internal runs; mean_score; uncertainty is standard error, not a confidence interval.",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "FrontierMath tiers 1–3 v2",
      "model": "claude-opus-5-5",
      "value": 91.22807017543859,
      "unit": "percent",
      "configuration": "max",
      "interval": "SE 1.6786209858442096",
      "notes": "Epoch AI internal runs; mean_score; uncertainty is standard error, not a confidence interval.",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "FrontierMath tiers 1–3 v2",
      "model": "claude-sonnet-5-5",
      "value": 88.7719298245614,
      "unit": "percent",
      "configuration": "max",
      "interval": "SE 1.8734030360524248",
      "notes": "Epoch AI internal runs; mean_score; uncertainty is standard error, not a confidence interval.",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "FrontierMath tier 4 v2",
      "model": "gpt-6-astra",
      "value": 97.6,
      "unit": "percent",
      "configuration": "max",
      "interval": "SE 2.4",
      "notes": "Epoch AI internal runs; mean_score; uncertainty is standard error, not a confidence interval.",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "FrontierMath tier 4 v2",
      "model": "gpt-6-1-sol",
      "value": 100.0,
      "unit": "percent",
      "configuration": "max",
      "interval": "SE 0.0",
      "notes": "Epoch AI internal runs; mean_score; uncertainty is standard error, not a confidence interval.",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "FrontierMath tier 4 v2",
      "model": "claude-opus-5-5",
      "value": 95.0,
      "unit": "percent",
      "configuration": "max",
      "interval": "SE 4.117547077105885",
      "notes": "Epoch AI internal runs; mean_score; uncertainty is standard error, not a confidence interval.",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "FrontierMath tier 4 v2",
      "model": "claude-sonnet-5-5",
      "value": 80.48780487804879,
      "unit": "percent",
      "configuration": "max",
      "interval": "SE 6.265967111543964",
      "notes": "Epoch AI internal runs; mean_score; uncertainty is standard error, not a confidence interval.",
      "source": "https://epoch.ai/benchmarks/use-this-data",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "Epoch Capabilities Index",
      "model": "gpt-6-astra",
      "value": 168.81,
      "unit": "ECI",
      "configuration": "composite across benchmark configurations",
      "interval": "165.45–174.51",
      "notes": "Epoch AI, Capabilities & benchmarking, CC BY. Downloaded 2026-10-24; not a matched-effort test.",
      "source": "https://epoch.ai/eci",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "Epoch Capabilities Index",
      "model": "gpt-6-1-sol",
      "value": 168.37,
      "unit": "ECI",
      "configuration": "composite across benchmark configurations",
      "interval": "165.0–174.43",
      "notes": "Epoch AI, Capabilities & benchmarking, CC BY. Downloaded 2026-10-24; not a matched-effort test.",
      "source": "https://epoch.ai/eci",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "Epoch Capabilities Index",
      "model": "claude-opus-5-5",
      "value": 168.83,
      "unit": "ECI",
      "configuration": "composite across benchmark configurations",
      "interval": "165.25–173.95",
      "notes": "Epoch AI, Capabilities & benchmarking, CC BY. Downloaded 2026-10-24; not a matched-effort test.",
      "source": "https://epoch.ai/eci",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "Epoch AI",
      "metric": "Epoch Capabilities Index",
      "model": "claude-sonnet-5-5",
      "value": 166.18,
      "unit": "ECI",
      "configuration": "composite across benchmark configurations",
      "interval": "163.06–171.34",
      "notes": "Epoch AI, Capabilities & benchmarking, CC BY. Downloaded 2026-10-24; not a matched-effort test.",
      "source": "https://epoch.ai/eci",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-1 standard",
      "model": "gpt-6-astra",
      "value": 97.5,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-1 standard",
      "model": "gpt-6-1-sol",
      "value": 96.5,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-1 standard",
      "model": "claude-opus-5-5",
      "value": 97.5,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-1 standard",
      "model": "claude-sonnet-5-5",
      "value": 96.5,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-2 standard",
      "model": "gpt-6-astra",
      "value": 95.0,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-2 standard",
      "model": "gpt-6-1-sol",
      "value": 94.16666666666667,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-2 standard",
      "model": "claude-opus-5-5",
      "value": 91.66666666666666,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-2 standard",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-3 standard",
      "model": "gpt-6-astra",
      "value": 62.71280210060628,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-3 standard",
      "model": "gpt-6-1-sol",
      "value": 52.72904006820879,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-3 standard",
      "model": "claude-opus-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-3 standard",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-1 provider adapter",
      "model": "gpt-6-astra",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-1 provider adapter",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-1 provider adapter",
      "model": "claude-opus-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-1 provider adapter",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-2 provider adapter",
      "model": "gpt-6-astra",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-2 provider adapter",
      "model": "gpt-6-1-sol",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-2 provider adapter",
      "model": "claude-opus-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-2 provider adapter",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-3 provider adapter",
      "model": "gpt-6-astra",
      "value": 98.55074288592294,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-3 provider adapter",
      "model": "gpt-6-1-sol",
      "value": 96.17739071933946,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-3 provider adapter",
      "model": "claude-opus-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    },
    {
      "publisher": "ARC Prize",
      "metric": "ARC-AGI-3 provider adapter",
      "model": "claude-sonnet-5-5",
      "value": null,
      "unit": "percent",
      "configuration": "max-provider-adapter",
      "interval": null,
      "notes": "Semi-private; model + harness result. Missing is not zero. Adapter route is a different configuration.",
      "source": "https://arcprize.org/leaderboard",
      "retrievedAt": "2026-10-24"
    }
  ]
}