{
  "api_version": "1",
  "updated": "2026-10-26",
  "latest_event": "2026-10-24",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "claude-opus-5",
    "name": "Claude Opus 5",
    "lab": "anthropic",
    "date": "2026-07-24",
    "flagship": true,
    "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
    "releases": [],
    "variant_of": null,
    "made_by": "Anthropic",
    "url": "https://frontierlog.ai/model/claude-opus-5",
    "results": [
      {
        "id": "claude-opus-5:hle:no-tools:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 56.3,
        "unit": "%",
        "printed": "56.3%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/news/claude-opus-5",
        "source_title": "Introducing Claude Opus 5",
        "source_date": "2026-07-24",
        "where": "benchmark table (image), row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Opus 5', 'no tools'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-16",
        "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": "56.3",
            "where": "Table 8.1.A (p.152), row Humanity's Last Exam / No tools, column Opus 5"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260910/",
            "name": "DeepSeek-V4.1-Flash",
            "title": "DeepSeek-V4.1-Flash: Smarter, Faster, More Efficient",
            "lab": "deepseek",
            "date": "2026-09-10",
            "printed": "56.3",
            "where": "Benchmark table image; row 'HLE'; column 'Claude Opus 5'; separate 'HLE (w/tools)' row exists"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news261017/",
            "name": "DeepSeek-V4.1-Flash-1017",
            "title": "DeepSeek-V4.1-Flash-1017: Wish-Driven Programming Made Real",
            "lab": "deepseek",
            "date": "2026-10-17",
            "printed": "56.3",
            "where": "Benchmark table image; row 'HLE (w/o tools)'; column 'Claude Opus 5'; first value of 'x/y*' cells is the full set"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
      },
      {
        "id": "claude-opus-5:hle:no-tools:2",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 56.6,
        "unit": "%",
        "printed": "56.6",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Claude Opus 5",
        "source": "https://www.anthropic.com/claude-opus-5-5-system-card",
        "source_title": "Claude Opus 5.5 System Card",
        "source_date": "2026-09-22",
        "where": "Section 8.1 evaluation summary table, row 'Humanity's Last Exam, No tools', column 'Claude Opus 5'",
        "own": false,
        "primary": false,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-16",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system-card",
            "name": "Claude Fable 5.1 and Mythos 5.1 card",
            "title": "Claude Fable 5.1 and Claude Mythos 5.1 System Card",
            "lab": "anthropic",
            "date": "2026-09-01",
            "printed": "56.6%",
            "where": "Table 8.1.A (p.167), row Humanity's Last Exam / No tools, column Claude Opus 5"
          },
          {
            "url": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
            "name": "Claude Fable 5.1 and Mythos 5.1",
            "title": "Introducing Claude Fable 5.1 and Claude Mythos 5.1",
            "lab": "anthropic",
            "date": "2026-09-01",
            "printed": "56.6%",
            "where": "HTML benchmark table 'Here, you can see how Fable 5.1 compares', row Humanity's Last Exam (no tools), column Opus 5"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/8a49bd8c526c6c66.png"
      },
      {
        "id": "claude-opus-5:hle:with-tools:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 64.7,
        "unit": "%",
        "printed": "64.7%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/news/claude-opus-5",
        "source_title": "Introducing Claude Opus 5",
        "source_date": "2026-07-24",
        "where": "benchmark table (image), row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Opus 5', 'with tools'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "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": "64.7",
            "where": "Table 8.1.A (p.152), row Humanity's Last Exam / With tools, column Opus 5"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
      },
      {
        "id": "claude-opus-5:arc-agi-2:default:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "arc-agi-2",
        "benchmark_name": "ARC-AGI-2",
        "setting": "default",
        "setting_name": "",
        "score": 90.4,
        "unit": "%",
        "printed": "90.4",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/claude-opus-5-system-card",
        "source_title": "Claude Opus 5 System Card",
        "source_date": "2026-07-24",
        "where": "Section 8.1 capability evaluation summary table (page 153), row 'ARC-AGI-2', column 'Opus 5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "max effort; semi-private set; as reported by the ARC Prize Foundation",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system-card",
            "name": "Claude Fable 5.1 and Mythos 5.1 card",
            "title": "Claude Fable 5.1 and Claude Mythos 5.1 System Card",
            "lab": "anthropic",
            "date": "2026-09-01",
            "printed": "90.42%",
            "where": "Table 8.1.A (p.167), row ARC-AGI-2, column Claude Opus 5"
          },
          {
            "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": "90.4%",
            "where": "table 8 \"ARC-AGI-2\" row, \"Claude Opus 5\" column"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/f64941f07dfe6a79.png"
      },
      {
        "id": "claude-opus-5:hle:with-tools:2",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 63.6,
        "unit": "%",
        "printed": "63.6%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
        "source_title": "Introducing Claude Fable 5.1 and Claude Mythos 5.1",
        "source_date": "2026-09-01",
        "where": "benchmark table (HTML), row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Opus 5', 'with tools'",
        "own": false,
        "primary": false,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system-card",
            "name": "Claude Fable 5.1 and Mythos 5.1 card",
            "title": "Claude Fable 5.1 and Claude Mythos 5.1 System Card",
            "lab": "anthropic",
            "date": "2026-09-01",
            "printed": "63.6%",
            "where": "Table 8.1.A (p.167), row Humanity's Last Exam / With tools, column Claude Opus 5"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-5-5-system-card",
            "name": "Claude Opus 5.5 card",
            "title": "Claude Opus 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "63.6",
            "where": "Table 8.1.A (p.174), row Humanity's Last Exam / With tools, column Claude Opus 5"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-5-5",
            "name": "Claude Opus 5.5",
            "title": "Introducing Claude Opus 5.5",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "63.6%",
            "where": "HTML benchmark table under 'Performance and cost-effectiveness', row Humanity's Last Exam (with tools), column Opus 5"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260910/",
            "name": "DeepSeek-V4.1-Flash",
            "title": "DeepSeek-V4.1-Flash: Smarter, Faster, More Efficient",
            "lab": "deepseek",
            "date": "2026-09-10",
            "printed": "63.6",
            "where": "Benchmark table image; row 'HLE (w/tools)'; column 'Claude Opus 5'"
          },
          {
            "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": "63.6%",
            "where": "table 3 \"Humanity's Last Exam (w/ tools)\" row, \"Claude Opus 5\" column"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/f381a9ad69b68272.png"
      },
      {
        "id": "claude-opus-5:browsecomp:default:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "browsecomp",
        "benchmark_name": "BrowseComp",
        "setting": "default",
        "setting_name": "",
        "score": 90.8,
        "unit": "%",
        "printed": "90.8%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/news/claude-opus-5",
        "source_title": "Introducing Claude Opus 5",
        "source_date": "2026-07-24",
        "where": "benchmark table (image), row 'Agentic search, BrowseComp', column 'Opus 5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "compaction at 200K tokens",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "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": "90.8",
            "where": "Table 8.1.A (p.152), row BrowseComp, column Opus 5"
          },
          {
            "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": "90.8%",
            "where": "table 1 \"BrowseComp\" row, \"Claude Opus 5\" column"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
      },
      {
        "id": "claude-opus-5:swe-bench-pro:default:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "swe-bench-pro",
        "benchmark_name": "SWE-bench Pro",
        "setting": "default",
        "setting_name": "",
        "score": 79.2,
        "unit": "%",
        "printed": "79.2",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/claude-opus-5-system-card",
        "source_title": "Claude Opus 5 System Card",
        "source_date": "2026-07-24",
        "where": "Section 8.1 Evaluation summary, Table 8.1.A, row 'SWE-bench Pro', column 'Opus 5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "adaptive thinking, max effort; average of 5 trials",
        "caveat": "The source doesn't say whether this is the public or the commercial set",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system-card",
            "name": "Claude Fable 5.1 and Mythos 5.1 card",
            "title": "Claude Fable 5.1 and Claude Mythos 5.1 System Card",
            "lab": "anthropic",
            "date": "2026-09-01",
            "printed": "79.2",
            "where": "Table 8.1.A (p.167), row SWE-bench Pro, column Claude Opus 5"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-5-5-system-card",
            "name": "Claude Opus 5.5 card",
            "title": "Claude Opus 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "79.2",
            "where": "Table 8.1.A (p.174), row SWE-bench Pro, column Claude Opus 5"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/057c8f7b933f9831.png"
      },
      {
        "id": "claude-opus-5:deepswe-1-1:default:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "deepswe-1-1",
        "benchmark_name": "DeepSWE v1.1",
        "setting": "default",
        "setting_name": "",
        "score": 68.8,
        "unit": "%",
        "printed": "68.8%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/news/claude-opus-5",
        "source_title": "Introducing Claude Opus 5",
        "source_date": "2026-07-24",
        "where": "benchmark table (image), row 'Agentic coding, DeepSWE v1.1', column 'Opus 5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "average of 5 trials",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "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": "68.8",
            "where": "Table 8.1.A (p.152), row DeepSWE v1.1, column Opus 5"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
      },
      {
        "id": "claude-opus-5:terminal-bench-4-0:default:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "terminal-bench-4-0",
        "benchmark_name": "Terminal-Bench 4.0",
        "setting": "default",
        "setting_name": "",
        "score": 52.3,
        "unit": "%",
        "printed": "52.3%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
        "source_title": "Introducing Claude Fable 5.1 and Claude Mythos 5.1",
        "source_date": "2026-09-01",
        "where": "benchmark table, row 'Terminal-Bench 4.0', column 'Opus 5'",
        "own": false,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system-card",
            "name": "Claude Fable 5.1 and Mythos 5.1 card",
            "title": "Claude Fable 5.1 and Claude Mythos 5.1 System Card",
            "lab": "anthropic",
            "date": "2026-09-01",
            "printed": "52.3%",
            "where": "Section 8.6 (p.171) sentence: \"Claude Opus 5 scored 52.3%\""
          },
          {
            "url": "https://www.anthropic.com/claude-opus-5-5-system-card",
            "name": "Claude Opus 5.5 card",
            "title": "Claude Opus 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "52.3",
            "where": "Table 8.1.A (p.174), row Terminal-Bench 4.0, column Claude Opus 5"
          },
          {
            "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card",
            "name": "Claude Sonnet 5.5 card",
            "title": "Claude Sonnet 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-28",
            "printed": "52.3%",
            "where": "Section 8.5 (p.113) sentence: \"and Claude Opus 5 scored 52.3%\""
          },
          {
            "url": "https://www.anthropic.com/claude-opus-5-5",
            "name": "Claude Opus 5.5",
            "title": "Introducing Claude Opus 5.5",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "52.3%",
            "where": "HTML benchmark table under 'Performance and cost-effectiveness', row Terminal-Bench 4.0¹, column Opus 5"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/f381a9ad69b68272.png"
      },
      {
        "id": "claude-opus-5:osworld-2-0:offline-partial:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "osworld-2-0",
        "benchmark_name": "OSWorld 2.0",
        "setting": "offline-partial",
        "setting_name": "Offline set, partial score",
        "score": 70.6,
        "unit": "%",
        "printed": "70.6%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/news/claude-opus-5",
        "source_title": "Introducing Claude Opus 5",
        "source_date": "2026-07-24",
        "where": "benchmark table image, row 'Computer use / OSWorld 2.0', column 'Opus 5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "task release not stated",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-18",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": [],
        "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
      },
      {
        "id": "claude-opus-5:arc-agi-3:default:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "arc-agi-3",
        "benchmark_name": "ARC-AGI-3",
        "setting": "default",
        "setting_name": "",
        "score": 30.2,
        "unit": "%",
        "printed": "30.2%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/news/claude-opus-5",
        "source_title": "Introducing Claude Opus 5",
        "source_date": "2026-07-24",
        "where": "benchmark table image, row 'Novel problem-solving / ARC-AGI-3', column 'Opus 5'",
        "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": "30.2",
            "where": "Table 8.1.A (p.153), row ARC-AGI-3, column Opus 5 (cell \"30.2 (high)\")"
          },
          {
            "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": "30.2%",
            "where": "table 8 \"ARC-AGI-3\" row, \"Claude Opus 5\" column"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
      },
      {
        "id": "claude-opus-5:osworld-2-0:offline-partial:2",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "osworld-2-0",
        "benchmark_name": "OSWorld 2.0",
        "setting": "offline-partial",
        "setting_name": "Offline set, partial score",
        "score": 75.4,
        "unit": "%",
        "printed": "75.4%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
        "source_title": "Introducing Claude Fable 5.1 and Claude Mythos 5.1",
        "source_date": "2026-09-01",
        "where": "benchmark table (HTML), row 'Computer use / OSWorld 2.0 [2]', column 'Opus 5', the value tagged 'partial'",
        "own": false,
        "primary": false,
        "parallel": false,
        "notes": "partial score; August 2026 task release, not comparable to earlier OSWorld 2.0 scores",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-19",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system-card",
            "name": "Claude Fable 5.1 and Mythos 5.1 card",
            "title": "Claude Fable 5.1 and Claude Mythos 5.1 System Card",
            "lab": "anthropic",
            "date": "2026-09-01",
            "printed": "75.4%",
            "where": "section 8.14.3 sentence: 'Claude Opus 5 scored 75.4% partial' on the August 2026 task release (Table 8.1.A OSWorld 2.0 (partial/strict) prints 75.4/39.6)"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/f381a9ad69b68272.png"
      },
      {
        "id": "claude-opus-5:automationbench:default:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "automationbench",
        "benchmark_name": "AutomationBench",
        "setting": "default",
        "setting_name": "",
        "score": 26.0,
        "unit": "%",
        "printed": "26.0%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/news/claude-opus-5",
        "source_title": "Introducing Claude Opus 5",
        "source_date": "2026-07-24",
        "where": "benchmark table (image), row 'Business workflows, AutomationBench', column 'Opus 5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "max effort; Zapier leaderboard, private held-out set",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "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": "26.0",
            "where": "Table 8.1.A, row AutomationBench, column Opus 5 (also 8.13.7 text: 'Claude Opus 5 (max effort) scored 26.0%')"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
      },
      {
        "id": "claude-opus-5:automationbench:default:2",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "automationbench",
        "benchmark_name": "AutomationBench",
        "setting": "default",
        "setting_name": "",
        "score": 24.0,
        "unit": "%",
        "printed": "24%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Claude Opus 5 (medium effort)",
        "source": "https://www.anthropic.com/claude-opus-5-system-card",
        "source_title": "Claude Opus 5 System Card",
        "source_date": "2026-07-24",
        "where": "Section 8.13.7 'AutomationBench', paragraph after the benchmark description, the medium-effort figure",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "medium effort; Zapier leaderboard, private held-out set",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-20",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "claude-opus-5:automationbench:default:3",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "automationbench",
        "benchmark_name": "AutomationBench",
        "setting": "default",
        "setting_name": "",
        "score": 26.9,
        "unit": "%",
        "printed": "26.9%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/claude-fable-and-mythos-5-1",
        "source_title": "Introducing Claude Fable 5.1 and Claude Mythos 5.1",
        "source_date": "2026-09-01",
        "where": "benchmark table (HTML), row 'Business workflows, AutomationBench', column 'Opus 5'",
        "own": false,
        "primary": false,
        "parallel": false,
        "notes": "Zapier leaderboard, private held-out set; max effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system-card",
            "name": "Claude Fable 5.1 and Mythos 5.1 card",
            "title": "Claude Fable 5.1 and Claude Mythos 5.1 System Card",
            "lab": "anthropic",
            "date": "2026-09-01",
            "printed": "26.9",
            "where": "Table 8.1.A, row AutomationBench, column Claude Opus 5 (also 8.15.6 text: 'Claude Opus 5 at max effort at 26.9%')"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-5-5-system-card",
            "name": "Claude Opus 5.5 card",
            "title": "Claude Opus 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "26.9",
            "where": "Table 8.1.A, row AutomationBench, column Claude Opus 5 (also 8.14.6 text: 'Claude Opus 5 at 26.9%')"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-5-5",
            "name": "Claude Opus 5.5",
            "title": "Introducing Claude Opus 5.5",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "26.9%",
            "where": "Benchmark table, row Business workflows AutomationBench, column Opus 5"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/f381a9ad69b68272.png"
      },
      {
        "id": "claude-opus-5:frontiercode-1-1:main:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "frontiercode-1-1",
        "benchmark_name": "FrontierCode 1.1",
        "setting": "main",
        "setting_name": "Main set",
        "score": 53.4,
        "unit": "%",
        "printed": "53.4%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/news/claude-opus-5",
        "source_title": "Introducing Claude Opus 5",
        "source_date": "2026-07-24",
        "where": "benchmark table (image), row 'Agentic coding, FrontierCode v1.1, Main', column 'Opus 5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "run and scored by Cognition; medium effort, the best of five effort levels; mean of 5 runs",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "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": "53.4",
            "where": "Table 8.1.A, row FrontierCode 1.1 (Main), column Opus 5; 8.4 text: '53.4% score (each model at its best reasoning effort)', Figure 8.4.A caption: best at medium effort"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-5-5-system-card",
            "name": "Claude Opus 5.5 card",
            "title": "Claude Opus 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "53.4%",
            "where": "Section 8.4 text: 'improving on Claude Opus 5 (53.4%)' (each model at its best reasoning effort)"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
      },
      {
        "id": "claude-opus-5:frontiercode-1-1:extended:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "frontiercode-1-1",
        "benchmark_name": "FrontierCode 1.1",
        "setting": "extended",
        "setting_name": "Extended set",
        "score": 63.6,
        "unit": "%",
        "printed": "63.6%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Claude Opus 5",
        "source": "https://www.anthropic.com/claude-opus-5-system-card",
        "source_title": "Claude Opus 5 System Card",
        "source_date": "2026-07-24",
        "where": "Section 8.4 'FrontierCode', paragraph ranking Opus 5, sentence on FrontierCode (Extended)",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "run and scored by Cognition; medium effort, the best of five effort levels; mean of 5 runs",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-5-5-system-card",
            "name": "Claude Opus 5.5 card",
            "title": "Claude Opus 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "63.6%",
            "where": "Section 8.4 text: 'improving on Claude Opus 5 (63.6%)' on FrontierCode (Extended), each model at its best reasoning effort"
          }
        ]
      },
      {
        "id": "claude-opus-5:frontiercode-1-1:main:2",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "frontiercode-1-1",
        "benchmark_name": "FrontierCode 1.1",
        "setting": "main",
        "setting_name": "Main set",
        "score": 48.0,
        "unit": "%",
        "printed": "48.0%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/claude-opus-5-5",
        "source_title": "Introducing Claude Opus 5.5",
        "source_date": "2026-09-22",
        "where": "benchmark table (HTML), row 'Agentic coding, FrontierCode v1.1 (Main)', column 'Opus 5'",
        "own": false,
        "primary": false,
        "parallel": false,
        "notes": "run by Cognition in Claude Code",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-5-5-system-card",
            "name": "Claude Opus 5.5 card",
            "title": "Claude Opus 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "48.0",
            "where": "Table 8.1.A, row FrontierCode v1.1 (Main), column Claude Opus 5 (Claude models run in Claude Code per Figure 8.4.A caption)"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/ca3ad55d7ec61f3c.png"
      },
      {
        "id": "claude-opus-5:osworld-2-1:partial:1",
        "model": "claude-opus-5",
        "model_name": "Claude Opus 5",
        "model_date": "2026-07-24",
        "event": "https://frontierlog.ai/event/2026-07-24-claude-opus-5",
        "benchmark": "osworld-2-1",
        "benchmark_name": "OSWorld 2.1",
        "setting": "partial",
        "setting_name": "Full set, partial score",
        "score": 74.0,
        "unit": "%",
        "printed": "74.0%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 5",
        "source": "https://www.anthropic.com/claude-opus-5-5",
        "source_title": "Introducing Claude Opus 5.5",
        "source_date": "2026-09-22",
        "where": "benchmark table (HTML), row 'Computer use / OSWorld 2.1', column 'Opus 5', the value tagged 'partial'",
        "own": false,
        "primary": true,
        "parallel": false,
        "notes": "max effort; mean of 5 runs; September 10, 2026 task files; the system card labels it OSWorld 2.0",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-20",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-5-5-system-card",
            "name": "Claude Opus 5.5 card",
            "title": "Claude Opus 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-09-22",
            "printed": "74.0%",
            "where": "section 8.13.3 OSWorld 2.0 (p. 206), sentence: 'Claude Opus 5 scored 74.0% partial and 37.2% strict pass'; also 8.1 Evaluation summary, row 'OSWorld 2.0 (partial/strict)', column Claude Opus 5: '74.0/37.2'"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/ca3ad55d7ec61f3c.png"
      }
    ]
  }
}