{
  "api_version": "1",
  "updated": "2026-11-10",
  "latest_event": "2026-11-09",
  "rights": "© Frontier Log, all rights reserved.",
  "benchmark": {
    "slug": "mcp-atlas",
    "name": "MCP Atlas",
    "family": "mcp-atlas",
    "unit": "%",
    "higher_is_better": true,
    "status": "retired",
    "active": false,
    "superseded_by": null,
    "retired": "2026-07-24",
    "benchmark_url": "https://labs.scale.com/leaderboard/mcp_atlas",
    "introduced": "2025-09-19",
    "creator": "Scale AI",
    "introduced_source": "https://scale.com/blog/mcp-atlas",
    "settings": [
      {
        "setting": "public",
        "name": "Public set (500 tasks)",
        "url": "https://frontierlog.ai/benchmark/mcp-atlas",
        "results": 26
      },
      {
        "setting": "full",
        "name": "Full set (1,000 tasks)",
        "url": "https://frontierlog.ai/benchmark/mcp-atlas/full",
        "results": 8
      },
      {
        "setting": "unstated",
        "name": "Set not stated",
        "url": "https://frontierlog.ai/benchmark/mcp-atlas/unstated",
        "results": 23
      }
    ]
  },
  "total": 23,
  "results": [
    {
      "id": "gpt-5-mini:mcp-atlas:unstated:1",
      "model": "gpt-5-mini",
      "model_name": "GPT-5 mini",
      "model_date": "2025-08-07",
      "event": "https://frontierlog.ai/event/2025-08-07-gpt-5",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 47.6,
      "unit": "%",
      "printed": "47.6%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5 mini (high)",
      "source": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
      "source_title": "Introducing GPT-5.4 mini and nano",
      "source_date": "2026-03-17",
      "where": "detailed results, 'Tool-calling' table (HTML; columns GPT-5.4 (xhigh), GPT-5.4 mini (xhigh), GPT-5.4 nano (xhigh), GPT-5 mini (high¹)), row 'MCP Atlas', column 'GPT-5 mini (high¹)'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "high reasoning effort, the highest available",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-07",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/7787ac3e92d17a83.png"
    },
    {
      "id": "gpt-5-1:mcp-atlas:unstated:1",
      "model": "gpt-5-1",
      "model_name": "GPT-5.1",
      "model_date": "2025-11-12",
      "event": "https://frontierlog.ai/event/2025-11-12-gpt-5-1",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 44.5,
      "unit": "%",
      "printed": "44.5%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5.1 Thinking",
      "source": "https://openai.com/index/introducing-gpt-5-2",
      "source_title": "Introducing GPT-5.2",
      "source_date": "2025-12-11",
      "where": "tool usage eval table (HTML; columns GPT-5.2 Thinking, GPT-5.2 Pro, GPT-5.1 Thinking), row 'Scale MCP-Atlas', column 'GPT-5.1 Thinking'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "high reasoning effort",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-29",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/0e9d54849c414ffc.png"
    },
    {
      "id": "claude-opus-4-5:mcp-atlas:unstated:1",
      "model": "claude-opus-4-5",
      "model_name": "Claude Opus 4.5",
      "model_date": "2025-11-24",
      "event": "https://frontierlog.ai/event/2025-11-24-claude-opus-4-5",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 62.3,
      "unit": "%",
      "printed": "62.3%",
      "reported_by": "anthropic",
      "run_by": null,
      "label": "Opus 4.5",
      "source": "https://www.anthropic.com/news/claude-opus-4-5",
      "source_title": "Introducing Claude Opus 4.5",
      "source_date": "2025-11-24",
      "where": "benchmark table (image), row 'Scaled tool use, MCP Atlas', column 'Opus 4.5'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "no extended thinking; 200K context",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://www.anthropic.com/claude-opus-4-6-system-card",
          "name": "Claude Opus 4.6 card",
          "title": "Claude Opus 4.6 System Card",
          "lab": "anthropic",
          "date": "2026-02-05",
          "printed": "62.3%",
          "where": "Table 2.3.A, row MCP-Atlas, column Claude Opus 4.5; Section 2.16 \"Claude Opus 4.5's 62.3%\""
        },
        {
          "url": "https://www.anthropic.com/claude-opus-4-5-system-card",
          "name": "Claude Opus 4.5 card",
          "title": "Claude Opus 4.5 System Card",
          "lab": "anthropic",
          "date": "2025-11-24",
          "printed": "62.3%",
          "where": "Table 2.3.A row MCP Atlas (footnote 4: without extended thinking); Section 2.12 \"Claude Opus 4.5 scored 62.3% on MCP-Atlas\""
        },
        {
          "url": "https://www.anthropic.com/claude-sonnet-4-6-system-card",
          "name": "Claude Sonnet 4.6 card",
          "title": "Claude Sonnet 4.6 System Card",
          "lab": "anthropic",
          "date": "2026-02-17",
          "printed": "62.3%",
          "where": "Summary table row MCP-Atlas, column Opus 4.5; Section MCP-Atlas \"top score of 62.3% by Claude Opus 4.5\""
        },
        {
          "url": "https://www.anthropic.com/news/claude-opus-4-6",
          "name": "Claude Opus 4.6",
          "title": "Claude Opus 4.6",
          "lab": "anthropic",
          "date": "2026-02-05",
          "printed": "62.3%",
          "where": "Benchmark table image, row Scaled tool use / MCP Atlas, column Opus 4.5"
        },
        {
          "url": "https://www.anthropic.com/news/claude-sonnet-4-6",
          "name": "Claude Sonnet 4.6",
          "title": "Introducing Sonnet 4.6",
          "lab": "anthropic",
          "date": "2026-02-17",
          "printed": "62.3%",
          "where": "Benchmark table image, row MCP-Atlas, column Opus 4.5"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/790b0c5480216588.png",
      "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/52303b11db76017fd0c2f73c7fafa5c752515979-2600x2236.png"
    },
    {
      "id": "gpt-5-2:mcp-atlas:unstated:1",
      "model": "gpt-5-2",
      "model_name": "GPT-5.2",
      "model_date": "2025-12-11",
      "event": "https://frontierlog.ai/event/2025-12-11-gpt-5-2",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 60.6,
      "unit": "%",
      "printed": "60.6%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5.2 Thinking",
      "source": "https://openai.com/index/introducing-gpt-5-2",
      "source_title": "Introducing GPT-5.2",
      "source_date": "2025-12-11",
      "where": "tool usage eval table (HTML; columns GPT-5.2 Thinking, GPT-5.2 Pro, GPT-5.1 Thinking), row 'Scale MCP-Atlas', column 'GPT-5.2 Thinking'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "xhigh reasoning effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://openai.com/index/introducing-gpt-5-4",
          "name": "GPT-5.4",
          "title": "Introducing GPT-5.4",
          "lab": "openai",
          "date": "2026-03-05",
          "printed": "60.6%",
          "where": "Tool-use table, row 'MCP Atlas', column 'GPT‑5.2'"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/0e9d54849c414ffc.png"
    },
    {
      "id": "claude-opus-4-6:mcp-atlas:unstated:1",
      "model": "claude-opus-4-6",
      "model_name": "Claude Opus 4.6",
      "model_date": "2026-02-05",
      "event": "https://frontierlog.ai/event/2026-02-05-claude-opus-4-6",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 62.7,
      "unit": "%",
      "printed": "62.7%",
      "reported_by": "anthropic",
      "run_by": null,
      "label": "Claude Opus 4.6",
      "source": "https://www.anthropic.com/news/claude-opus-4-6",
      "source_title": "Claude Opus 4.6",
      "source_date": "2026-02-05",
      "where": "footnotes, MCP Atlas paragraph",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "high effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://www.anthropic.com/claude-opus-4-6-system-card",
          "name": "Claude Opus 4.6 card",
          "title": "Claude Opus 4.6 System Card",
          "lab": "anthropic",
          "date": "2026-02-05",
          "printed": "62.7%",
          "where": "Section 2.16: \"We obtained a score of 62.7% with high effort settings on Opus 4.6\"; Table 2.3.A footnote 3"
        }
      ]
    },
    {
      "id": "claude-opus-4-6:mcp-atlas:unstated:2",
      "model": "claude-opus-4-6",
      "model_name": "Claude Opus 4.6",
      "model_date": "2026-02-05",
      "event": "https://frontierlog.ai/event/2026-02-05-claude-opus-4-6",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 59.5,
      "unit": "%",
      "printed": "59.5%",
      "reported_by": "anthropic",
      "run_by": null,
      "label": "Opus 4.6",
      "source": "https://www.anthropic.com/news/claude-opus-4-6",
      "source_title": "Claude Opus 4.6",
      "source_date": "2026-02-05",
      "where": "benchmark table (image), row 'Scaled tool use, MCP Atlas', column 'Opus 4.6'",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "max effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://www.anthropic.com/claude-opus-4-6-system-card",
          "name": "Claude Opus 4.6 card",
          "title": "Claude Opus 4.6 System Card",
          "lab": "anthropic",
          "date": "2026-02-05",
          "printed": "59.5%",
          "where": "Section 2.16: \"Claude Opus 4.6 scored 59.5% on MCP-Atlas with max effort settings\"; Table 2.3.A"
        },
        {
          "url": "https://www.anthropic.com/claude-sonnet-4-6-system-card",
          "name": "Claude Sonnet 4.6 card",
          "title": "Claude Sonnet 4.6 System Card",
          "lab": "anthropic",
          "date": "2026-02-17",
          "printed": "59.5%",
          "where": "Summary table row MCP-Atlas, column Opus 4.6 (footnote 3, max effort)"
        },
        {
          "url": "https://www.anthropic.com/news/claude-sonnet-4-6",
          "name": "Claude Sonnet 4.6",
          "title": "Introducing Sonnet 4.6",
          "lab": "anthropic",
          "date": "2026-02-17",
          "printed": "59.5%",
          "where": "Benchmark table image, row Scaled tool use / MCP-Atlas, column Opus 4.6 (max effort per Opus 4.6 post)"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
      "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
    },
    {
      "id": "claude-sonnet-4-6:mcp-atlas:unstated:1",
      "model": "claude-sonnet-4-6",
      "model_name": "Claude Sonnet 4.6",
      "model_date": "2026-02-17",
      "event": "https://frontierlog.ai/event/2026-02-17-claude-sonnet-4-6",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 61.3,
      "unit": "%",
      "printed": "61.3%",
      "reported_by": "anthropic",
      "run_by": null,
      "label": "Sonnet 4.6",
      "source": "https://www.anthropic.com/news/claude-sonnet-4-6",
      "source_title": "Introducing Sonnet 4.6",
      "source_date": "2026-02-17",
      "where": "benchmark table (image), row 'Scaled tool use, MCP-Atlas', column 'Sonnet 4.6'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "max effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://www.anthropic.com/claude-sonnet-4-6-system-card",
          "name": "Claude Sonnet 4.6 card",
          "title": "Claude Sonnet 4.6 System Card",
          "lab": "anthropic",
          "date": "2026-02-17",
          "printed": "61.3%",
          "where": "Section MCP-Atlas: \"Claude Sonnet 4.6 scored 61.3% on MCP-Atlas with max effort settings\"; summary table"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/d694e6e89a414801.png",
      "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/10b2602771d21378cd6d76628a081c8a76dcf216-2600x2960.png"
    },
    {
      "id": "gpt-5-4:mcp-atlas:unstated:1",
      "model": "gpt-5-4",
      "model_name": "GPT-5.4",
      "model_date": "2026-03-05",
      "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 67.2,
      "unit": "%",
      "printed": "67.2%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.4",
      "source": "https://openai.com/index/introducing-gpt-5-4",
      "source_title": "Introducing GPT-5.4",
      "source_date": "2026-03-05",
      "where": "tool use eval table (HTML; columns GPT‑5.4, GPT‑5.4 Pro, GPT‑5.3-Codex, GPT‑5.2, GPT‑5.2 Pro), row 'MCP Atlas', column 'GPT‑5.4'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "xhigh reasoning effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
          "name": "GPT-5.4 mini and nano",
          "title": "Introducing GPT-5.4 mini and nano",
          "lab": "openai",
          "date": "2026-03-17",
          "printed": "67.2%",
          "where": "Tool-use table, row 'MCP Atlas', column 'GPT-5.4 (xhigh)'"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/762d23ad86ad7807.png"
    },
    {
      "id": "gpt-5-4:mcp-atlas:unstated:2",
      "model": "gpt-5-4",
      "model_name": "GPT-5.4",
      "model_date": "2026-03-05",
      "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 70.6,
      "unit": "%",
      "printed": "70.6%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.4",
      "source": "https://openai.com/index/introducing-gpt-5-5",
      "source_title": "Introducing GPT-5.5",
      "source_date": "2026-04-23",
      "where": "tool use eval table (HTML; columns GPT‑5.5, GPT‑5.4, GPT‑5.5 Pro, GPT‑5.4 Pro, Claude Opus 4.7, Gemini 3.1 Pro), row 'MCP Atlas**', column 'GPT‑5.4'",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "Scale AI results after its April 2026 update",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-29",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/1770b802461b91b1.png"
    },
    {
      "id": "gpt-5-4-mini:mcp-atlas:unstated:1",
      "model": "gpt-5-4-mini",
      "model_name": "GPT-5.4 mini",
      "model_date": "2026-03-17",
      "event": "https://frontierlog.ai/event/2026-03-17-gpt-5-4-mini-nano",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 57.7,
      "unit": "%",
      "printed": "57.7%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5.4 mini (xhigh)",
      "source": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
      "source_title": "Introducing GPT-5.4 mini and nano",
      "source_date": "2026-03-17",
      "where": "tool-calling table (HTML; columns GPT-5.4 (xhigh), GPT-5.4 mini (xhigh), GPT-5.4 nano (xhigh), GPT-5 mini (high¹)), row 'MCP Atlas', column 'GPT-5.4 mini (xhigh)'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "xhigh reasoning effort",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-29",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/7787ac3e92d17a83.png"
    },
    {
      "id": "gpt-5-4-nano:mcp-atlas:unstated:1",
      "model": "gpt-5-4-nano",
      "model_name": "GPT-5.4 nano",
      "model_date": "2026-03-17",
      "event": "https://frontierlog.ai/event/2026-03-17-gpt-5-4-mini-nano",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 56.1,
      "unit": "%",
      "printed": "56.1%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5.4 nano (xhigh)",
      "source": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
      "source_title": "Introducing GPT-5.4 mini and nano",
      "source_date": "2026-03-17",
      "where": "tool-calling table (HTML; columns GPT-5.4 (xhigh), GPT-5.4 mini (xhigh), GPT-5.4 nano (xhigh), GPT-5 mini (high¹)), row 'MCP Atlas', column 'GPT-5.4 nano (xhigh)'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "xhigh reasoning effort",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-29",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/7787ac3e92d17a83.png"
    },
    {
      "id": "claude-opus-4-7:mcp-atlas:unstated:1",
      "model": "claude-opus-4-7",
      "model_name": "Claude Opus 4.7",
      "model_date": "2026-04-16",
      "event": "https://frontierlog.ai/event/2026-04-16-claude-opus-4-7",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 79.7,
      "unit": "%",
      "printed": "79.7%",
      "reported_by": "anthropic",
      "run_by": "Scale AI",
      "label": "Claude Opus 4.7",
      "source": "https://www.anthropic.com/claude-opus-4-7-system-card",
      "source_title": "Claude Opus 4.7 System Card",
      "source_date": "2026-04-16",
      "where": "section on MCP-Atlas (p.210), text, extended config, the high-effort score",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "extended config of 256 turns and 100 tools; high effort",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-28",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "claude-opus-4-7:mcp-atlas:unstated:2",
      "model": "claude-opus-4-7",
      "model_name": "Claude Opus 4.7",
      "model_date": "2026-04-16",
      "event": "https://frontierlog.ai/event/2026-04-16-claude-opus-4-7",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 79.5,
      "unit": "%",
      "printed": "79.5%",
      "reported_by": "anthropic",
      "run_by": "Scale AI",
      "label": "Claude Opus 4.7",
      "source": "https://www.anthropic.com/claude-opus-4-7-system-card",
      "source_title": "Claude Opus 4.7 System Card",
      "source_date": "2026-04-16",
      "where": "section on MCP-Atlas (p.210), text, extended config, the max-effort score",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "extended config of 256 turns and 100 tools; max effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://www.anthropic.com/claude-opus-4-8-system-card",
          "name": "Claude Opus 4.8 card",
          "title": "Claude Opus 4.8 System Card",
          "lab": "anthropic",
          "date": "2026-05-28",
          "printed": "79.5%",
          "where": "Section MCP-Atlas: \"Opus 4.7 previously reported scores (79.5% for max) have changed as of April 2026\""
        }
      ]
    },
    {
      "id": "claude-opus-4-7:mcp-atlas:unstated:3",
      "model": "claude-opus-4-7",
      "model_name": "Claude Opus 4.7",
      "model_date": "2026-04-16",
      "event": "https://frontierlog.ai/event/2026-04-16-claude-opus-4-7",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 79.1,
      "unit": "%",
      "printed": "79.1",
      "reported_by": "anthropic",
      "run_by": "Scale AI",
      "label": "Claude Opus 4.7",
      "source": "https://www.anthropic.com/claude-opus-4-8-system-card",
      "source_title": "Claude Opus 4.8 System Card",
      "source_date": "2026-05-28",
      "where": "Table 8.1.A (p.195), row 'MCP-Atlas', column 'Claude Opus 4.7'",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "100-tool-call budget and updated judge",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-28",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "gpt-5-5:mcp-atlas: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": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 75.3,
      "unit": "%",
      "printed": "75.3%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.5",
      "source": "https://openai.com/index/introducing-gpt-5-5",
      "source_title": "Introducing GPT-5.5",
      "source_date": "2026-04-23",
      "where": "tool use eval table (HTML; columns GPT‑5.5, GPT‑5.4, GPT‑5.5 Pro, GPT‑5.4 Pro, Claude Opus 4.7, Gemini 3.1 Pro), row 'MCP Atlas**', column 'GPT‑5.5'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Scale AI results after its April 2026 update",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://z.ai/blog/glm-5.2",
          "name": "GLM-5.2",
          "title": "GLM-5.2: Built for Long-Horizon Tasks",
          "lab": "zai",
          "date": "2026-06-16",
          "printed": "75.3",
          "where": "benchmark table, MCP-Atlas (Public Set) row, GPT-5.5 column"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/1770b802461b91b1.png"
    },
    {
      "id": "claude-opus-4-8:mcp-atlas:unstated:1",
      "model": "claude-opus-4-8",
      "model_name": "Claude Opus 4.8",
      "model_date": "2026-05-28",
      "event": "https://frontierlog.ai/event/2026-05-28-claude-opus-4-8",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 82.2,
      "unit": "%",
      "printed": "82.2",
      "reported_by": "anthropic",
      "run_by": "Scale AI",
      "label": "Claude Opus 4.8",
      "source": "https://www.anthropic.com/claude-opus-4-8-system-card",
      "source_title": "Claude Opus 4.8 System Card",
      "source_date": "2026-05-28",
      "where": "Table 8.1.A (p.195), row 'MCP-Atlas', column 'Claude Opus 4.8'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "100-tool-call budget; adaptive thinking, max effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://www.anthropic.com/claude-fable-5-mythos-5-system-card",
          "name": "Claude Fable 5 and Mythos 5 card",
          "title": "Claude Fable 5 and Claude Mythos 5 System Card",
          "lab": "anthropic",
          "date": "2026-06-09",
          "printed": "82.2%",
          "where": "Section MCP-Atlas: \"Claude Fable 5 achieved an 83.3% pass rate, up from 82.2% for Claude Opus 4.8\""
        }
      ]
    },
    {
      "id": "claude-fable-5:mcp-atlas:unstated:1",
      "model": "claude-fable-5",
      "model_name": "Claude Fable 5",
      "model_date": "2026-06-09",
      "event": "https://frontierlog.ai/event/2026-06-09-claude-fable-5-mythos-5",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 83.3,
      "unit": "%",
      "printed": "83.3%",
      "reported_by": "anthropic",
      "run_by": null,
      "label": "Claude Fable 5",
      "source": "https://www.anthropic.com/claude-fable-5-mythos-5-system-card",
      "source_title": "Claude Fable 5 and Claude Mythos 5 System Card",
      "source_date": "2026-06-09",
      "where": "section 8.17.5 MCP Atlas (p.293), text",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-28",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "glm-5-2:mcp-atlas:unstated:1",
      "model": "glm-5-2",
      "model_name": "GLM-5.2",
      "model_date": "2026-06-16",
      "event": "https://frontierlog.ai/event/2026-06-16-glm-5-2",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 77.0,
      "unit": "%",
      "printed": "77.0",
      "reported_by": "zai",
      "run_by": null,
      "label": "GLM-5.2",
      "source": "https://z.ai/blog/glm-5.2",
      "source_title": "GLM-5.2: Built for Long-Horizon Tasks",
      "source_date": "2026-06-16",
      "where": "'LLM Performance Evaluation' chart image (eight panels), panel 'MCP-Atlas', first bar 'GLM-5.2', value label",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "maximum thinking effort; the post's chart, where its table prints 76.8",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/bfcfa600d8d45ed8.png",
      "source_image_original": "https://z-cdn-media.chatglm.cn/prompts-rich-media-resources/5.2-blog/20260617-012836.png"
    },
    {
      "id": "inkling:mcp-atlas:unstated:1",
      "model": "inkling",
      "model_name": "Inkling",
      "model_date": "2026-07-15",
      "event": "https://frontierlog.ai/event/2026-07-15-inkling",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 76.0,
      "unit": "%",
      "printed": "76.0%",
      "reported_by": "thinking-machines",
      "run_by": null,
      "label": "Inkling effort=0.99",
      "source": "https://thinkingmachines.ai/news/introducing-inkling/",
      "source_title": "Inkling: Our Open-Weights Model",
      "source_date": "2026-07-15",
      "where": "section 'Benchmarking Inkling', main benchmark table, row 'MCP Atlas' (Agentic (general)), column 'Inkling effort=0.99'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "effort 0.99",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-28",
      "corroborated_by": [
        {
          "url": "https://thinkingmachines.ai/news/inkling-small/",
          "name": "Thinking Machines post",
          "title": "Introducing Inkling-Small",
          "lab": "thinking-machines",
          "date": "2026-07-30",
          "printed": "76.0%",
          "where": "Full eval suite table, row MCP Atlas public / all, column Inkling, printed 78.8/76.0%: all split"
        },
        {
          "url": "https://research.meta.ai/blog/muse-spark-open",
          "name": "Muse Spark Open card",
          "title": "Introducing Muse Spark Open, The Most Capable US Open-Weight Model to Date",
          "lab": "meta",
          "date": "2026-10-15",
          "printed": "76.0",
          "where": "Benchmark scorecard image, row Agentic / MCP-Atlas, column Inkling (xhigh); another lab restating"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/a919a67037d3c8a6.png"
    },
    {
      "id": "inkling:mcp-atlas:unstated:2",
      "model": "inkling",
      "model_name": "Inkling",
      "model_date": "2026-07-15",
      "event": "https://frontierlog.ai/event/2026-07-15-inkling",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 74.1,
      "unit": "%",
      "printed": "74.1%",
      "reported_by": "thinking-machines",
      "run_by": null,
      "label": "Inkling",
      "source": "https://huggingface.co/thinkingmachines/Inkling",
      "source_title": "thinkingmachines/Inkling model card",
      "source_date": "2026-07-15",
      "where": "Hugging Face model card (launch revision 91b051f), section '5. Evaluations' table, row 'MCP Atlas', column 'Inkling'",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "effort 0.99",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-29",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/1763b2f8ffc4e612.png"
    },
    {
      "id": "claude-opus-5:mcp-atlas:unstated: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": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 85.8,
      "unit": "%",
      "printed": "85.8%",
      "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 8.13.2 MCP Atlas (p.181), text",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-28",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "longcat-2-5:mcp-atlas:unstated:1",
      "model": "longcat-2-5",
      "model_name": "LongCat-2.5",
      "model_date": "2026-11-01",
      "event": "https://frontierlog.ai/event/2026-11-01-longcat-2-5",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 83.1,
      "unit": "%",
      "printed": "83.1",
      "reported_by": "meituan",
      "run_by": null,
      "label": "LongCat-2.5",
      "source": "https://longcat.ai/blog/longcat-2.5",
      "source_title": "LongCat-2.5: Native Multimodality for Agentic Tasks",
      "source_date": "2026-11-01",
      "where": "'Evaluations' section table at the end of the post (rendered by the page's script; load https://longcat.ai/blog/longcat-2.5 in a browser and scroll to the end), row 'MCP Atlas' under 'General Agent', column 'LongCat-2.5'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "max reasoning effort; 256K context",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-02",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/2b178d4c19a6dbe7.png"
    },
    {
      "id": "ripple-65b-preview:mcp-atlas:unstated:1",
      "model": "ripple-65b-preview",
      "model_name": "Ripple-65B Preview",
      "model_date": "2026-11-04",
      "event": "https://frontierlog.ai/event/2026-11-04-ripple-65b-preview",
      "benchmark": "mcp-atlas",
      "benchmark_name": "MCP Atlas",
      "setting": "unstated",
      "setting_name": "Set not stated",
      "score": 71.67,
      "unit": "%",
      "printed": "71.67",
      "reported_by": "ricursive",
      "run_by": null,
      "label": "Ripple-65B Preview",
      "source": "https://huggingface.co/ricursiveAI/Ripple-65B-Preview",
      "source_title": "ricursiveAI/Ripple-65B-Preview model card",
      "source_date": "2026-11-04",
      "where": "Card, 'Instruct Model Benchmarks' table, row 'MCP Atlas', column 'Ripple-65B Preview'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "OpenCode harness; maximum effort",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-07",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/b689899dbfbbdf93.png"
    }
  ]
}