{
  "api_version": "1",
  "updated": "2026-11-10",
  "latest_event": "2026-11-09",
  "rights": "© Frontier Log, all rights reserved.",
  "benchmark": {
    "slug": "arc-agi-3",
    "name": "ARC-AGI-3",
    "family": "arc-agi",
    "unit": "%",
    "higher_is_better": true,
    "status": "active",
    "active": true,
    "superseded_by": null,
    "retired": null,
    "benchmark_url": "https://arcprize.org/arc-agi/3/",
    "introduced": "2026-03-25",
    "creator": "ARC Prize Foundation",
    "introduced_source": "https://arcprize.org/blog/arc-agi-3-launch",
    "settings": [
      {
        "setting": "semi-private",
        "name": "Semi-Private set",
        "url": "https://frontierlog.ai/benchmark/arc-agi-3",
        "results": 2
      },
      {
        "setting": "unstated",
        "name": "Set not stated",
        "url": "https://frontierlog.ai/benchmark/arc-agi-3/unstated",
        "results": 9
      }
    ]
  },
  "total": 9,
  "results": [
    {
      "id": "gpt-5-5:arc-agi-3:unstated:1",
      "model": "gpt-5-5",
      "model_name": "GPT-5.5",
      "model_date": "2026-04-23",
      "event": "https://frontierlog.ai/event/2026-04-23-gpt-5-5",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 0.43,
      "unit": "%",
      "printed": "0.43%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5.5",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "abstract-reasoning table (last table), row 'ARC-AGI-3⁷', column 'GPT-5.5'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-20",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/669da4e40aae1378.png"
    },
    {
      "id": "gpt-5-6-luna:arc-agi-3:unstated:1",
      "model": "gpt-5-6-luna",
      "model_name": "GPT-5.6 Luna",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 0.18,
      "unit": "%",
      "printed": "0.18%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.6 Luna",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "abstract-reasoning eval table (HTML, last table), row 'ARC-AGI-3⁷', column 'GPT‑5.6 Luna'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-26",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/91ab4e052a56be59.png"
    },
    {
      "id": "gpt-5-6-sol:arc-agi-3:unstated:1",
      "model": "gpt-5-6-sol",
      "model_name": "GPT-5.6 Sol",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 7.78,
      "unit": "%",
      "printed": "7.78%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.6 Sol",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "abstract-reasoning eval table (HTML, last table), row 'ARC-AGI-3⁷', column 'GPT‑5.6 Sol'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-18",
      "check": "human",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://www.anthropic.com/claude-opus-5-system-card",
          "name": "Claude Opus 5 card",
          "title": "Claude Opus 5 System Card",
          "lab": "anthropic",
          "date": "2026-07-24",
          "printed": "7.8",
          "where": "Table 8.1.A (p.153), row ARC-AGI-3, column GPT 5.6 Sol"
        },
        {
          "url": "https://www.anthropic.com/news/claude-opus-5",
          "name": "Claude Opus 5",
          "title": "Introducing Claude Opus 5",
          "lab": "anthropic",
          "date": "2026-07-24",
          "printed": "7.8%",
          "where": "benchmark table image, row ARC-AGI-3, column GPT-5.6 Sol"
        },
        {
          "url": "https://openai.com/index/gpt-6-astra",
          "name": "GPT-6 Astra",
          "title": "GPT-6 Astra: A new generation of intelligence",
          "lab": "openai",
          "date": "2026-09-03",
          "printed": "7.8%",
          "where": "table 8 \"ARC-AGI-3\" row, \"GPT-5.6 Sol\" column"
        },
        {
          "url": "https://openai.com/index/how-two-settings-tripled-our-arc-agi-3-scores",
          "name": "OpenAI post",
          "title": "How enabling two settings tripled our scores on the ARC-AGI-3 benchmark",
          "lab": "openai",
          "date": "2026-07-29",
          "printed": "7.8%",
          "where": "sentence: \"on ARC-AGI-3 ... GPT-5.6 Sol scored just 7.8%\""
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/669da4e40aae1378.png"
    },
    {
      "id": "gpt-5-6-terra:arc-agi-3:unstated:1",
      "model": "gpt-5-6-terra",
      "model_name": "GPT-5.6 Terra",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 0.8,
      "unit": "%",
      "printed": "0.8%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.6 Terra",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "abstract-reasoning eval table (HTML, last table), row 'ARC-AGI-3⁷', column 'GPT‑5.6 Terra'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-26",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/91ab4e052a56be59.png"
    },
    {
      "id": "gpt-6-astra:arc-agi-3:unstated:1",
      "model": "gpt-6-astra",
      "model_name": "GPT-6 Astra",
      "model_date": "2026-09-03",
      "event": "https://frontierlog.ai/event/2026-09-03-gpt-6-astra",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 99.9,
      "unit": "%",
      "printed": "99.9%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑6 Astra",
      "source": "https://openai.com/index/gpt-6-astra",
      "source_title": "GPT-6 Astra: A new generation of intelligence",
      "source_date": "2026-09-03",
      "where": "'Abstract reasoning' eval table (HTML), row 'ARC-AGI-3', column 'GPT‑6 Astra'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-18",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/6917e15279360166.png"
    },
    {
      "id": "claude-fable-5-2:arc-agi-3:unstated:1",
      "model": "claude-fable-5-2",
      "model_name": "Claude Fable 5.2",
      "model_date": "2026-11-09",
      "event": "https://frontierlog.ai/event/2026-11-09-claude-fable-5-2-and-mythos-5-2",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 79.26,
      "unit": "%",
      "printed": "79.26%",
      "reported_by": "anthropic",
      "run_by": "ARC Prize Foundation",
      "label": "Claude Fable 5.2",
      "source": "https://www.anthropic.com/claude-fable-5-2-mythos-5-2-system-card",
      "source_title": "Claude Fable 5.2 and Claude Mythos 5.2 System Card",
      "source_date": "2026-11-09",
      "where": "Section 8.15 'ARC-AGI-3', paragraph beginning 'The ARC Prize Foundation reports', Standard harness at high effort",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Standard harness; high effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "claude-fable-5-2:arc-agi-3:unstated:2",
      "model": "claude-fable-5-2",
      "model_name": "Claude Fable 5.2",
      "model_date": "2026-11-09",
      "event": "https://frontierlog.ai/event/2026-11-09-claude-fable-5-2-and-mythos-5-2",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 97.72,
      "unit": "%",
      "printed": "97.72%",
      "reported_by": "anthropic",
      "run_by": "ARC Prize Foundation",
      "label": "Claude Fable 5.2",
      "source": "https://www.anthropic.com/claude-fable-5-2-mythos-5-2-system-card",
      "source_title": "Claude Fable 5.2 and Claude Mythos 5.2 System Card",
      "source_date": "2026-11-09",
      "where": "Section 8.15 'ARC-AGI-3', paragraph beginning 'The ARC Prize Foundation reports', Provider Adapter harness at high effort",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "Provider Adapter harness; high effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "claude-fable-5-2:arc-agi-3:unstated:3",
      "model": "claude-fable-5-2",
      "model_name": "Claude Fable 5.2",
      "model_date": "2026-11-09",
      "event": "https://frontierlog.ai/event/2026-11-09-claude-fable-5-2-and-mythos-5-2",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 65.66,
      "unit": "%",
      "printed": "65.66%",
      "reported_by": "anthropic",
      "run_by": "ARC Prize Foundation",
      "label": "Fable 5.2",
      "source": "https://www.anthropic.com/claude-fable-5-2-mythos-5-2-system-card",
      "source_title": "Claude Fable 5.2 and Claude Mythos 5.2 System Card",
      "source_date": "2026-11-09",
      "where": "Section 8.15 'ARC-AGI-3', paragraph beginning 'The ARC Prize Foundation reports', Standard harness at max effort",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "Standard harness; max effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "claude-fable-5-2:arc-agi-3:unstated:4",
      "model": "claude-fable-5-2",
      "model_name": "Claude Fable 5.2",
      "model_date": "2026-11-09",
      "event": "https://frontierlog.ai/event/2026-11-09-claude-fable-5-2-and-mythos-5-2",
      "benchmark": "arc-agi-3",
      "benchmark_name": "ARC-AGI-3",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 94.9,
      "unit": "%",
      "printed": "94.90%",
      "reported_by": "anthropic",
      "run_by": "ARC Prize Foundation",
      "label": "Fable 5.2",
      "source": "https://www.anthropic.com/claude-fable-5-2-mythos-5-2-system-card",
      "source_title": "Claude Fable 5.2 and Claude Mythos 5.2 System Card",
      "source_date": "2026-11-09",
      "where": "Section 8.15 'ARC-AGI-3', paragraph beginning 'The ARC Prize Foundation reports', Provider Adapter harness at max effort",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "Provider Adapter harness; max effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    }
  ]
}