{
  "api_version": "1",
  "updated": "2026-10-26",
  "latest_event": "2026-10-24",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "claude-opus-4-6",
    "name": "Claude Opus 4.6",
    "lab": "anthropic",
    "date": "2026-02-05",
    "source": "https://www.anthropic.com/news/claude-opus-4-6",
    "source_title": "Claude Opus 4.6",
    "flagship": true,
    "event": "https://frontierlog.ai/event/2026-02-05-claude-opus-4-6",
    "releases": [],
    "variant_of": null,
    "made_by": "Anthropic",
    "url": "https://frontierlog.ai/model/claude-opus-4-6",
    "results": [
      {
        "id": "claude-opus-4-6:terminal-bench-2-0:default: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": "terminal-bench-2-0",
        "benchmark_name": "Terminal-Bench 2.0",
        "setting": "default",
        "setting_name": "",
        "score": 65.4,
        "unit": "%",
        "printed": "65.4%",
        "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 'Agentic terminal coding, Terminal-Bench 2.0', column 'Opus 4.6'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "Terminus 2 harness; 5 to 15 samples per task",
        "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-4-6-system-card",
            "name": "Claude Opus 4.6 card",
            "title": "Claude Opus 4.6 System Card",
            "lab": "anthropic",
            "date": "2026-02-05",
            "printed": "65.4%",
            "where": "Table 2.3.A (p.18), row Terminal-Bench 2.0, column Claude Opus 4.6"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "65.4%",
            "where": "Table 8.1.A (p.191), row Terminal-Bench 2.0, column Claude Opus 4.6"
          },
          {
            "url": "https://www.anthropic.com/news/claude-opus-4-7",
            "name": "Claude Opus 4.7",
            "title": "Introducing Claude Opus 4.7",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "65.4%",
            "where": "benchmark table image, row Terminal-Bench 2.0, column Opus 4.6"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "65.4",
            "where": "Bar chart image; group 'Terminal Bench 2.0 (Acc)'; value label on Claude-Opus-4.6-Max bar"
          },
          {
            "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
            "name": "Gemini 3.1 Pro",
            "title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
            "lab": "google",
            "date": "2026-02-19",
            "printed": "65.4%",
            "where": "Benchmark table image; row 'Terminal-Bench 2.0, Terminus-2 harness'; column claude-opus-4-6"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "65.4",
            "where": "Benchmark Table (appendix), row Terminal-Bench 2.0 (Terminus-2) ; column Claude Opus 4.6 (max effort)"
          },
          {
            "url": "https://qwen.ai/blog?id=qwen3.7",
            "name": "Qwen3.7-Max",
            "title": "Qwen3.7: The Agent Frontier",
            "lab": "alibaba",
            "date": "2026-05-20",
            "printed": "65.4",
            "where": "Score chart image (Qwen3.7-Max-Score.png), panel Terminal-Bench 2.0 (Terminus-2), value label on the Claude Opus-4.6 Max bar"
          },
          {
            "url": "https://z.ai/blog/glm-5.1",
            "name": "GLM-5.1",
            "title": "GLM-5.1: Towards Long-Horizon Tasks",
            "lab": "zai",
            "date": "2026-04-07",
            "printed": "65.4",
            "where": "Benchmark table, row 'Terminal-Bench 2.0 Terminus-2', column 'Claude Opus 4.6'"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
      },
      {
        "id": "claude-opus-4-6:swe-bench-verified:default: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": "swe-bench-verified",
        "benchmark_name": "SWE-bench Verified",
        "setting": "default",
        "setting_name": "",
        "score": 80.8,
        "unit": "%",
        "printed": "80.8%",
        "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 'Agentic coding, SWE-bench Verified', column 'Opus 4.6'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "averaged over 25 trials",
        "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-4-6-system-card",
            "name": "Claude Opus 4.6 card",
            "title": "Claude Opus 4.6 System Card",
            "lab": "anthropic",
            "date": "2026-02-05",
            "printed": "80.8%",
            "where": "Table 2.3.A (p.18), row SWE-bench Verified, column Claude Opus 4.6 (averaged over 25 trials)"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "80.8%",
            "where": "Table 8.1.A (p.191), row SWE-bench Verified, column Claude Opus 4.6"
          },
          {
            "url": "https://www.anthropic.com/news/claude-opus-4-7",
            "name": "Claude Opus 4.7",
            "title": "Introducing Claude Opus 4.7",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "80.8%",
            "where": "benchmark table image, row SWE-bench Verified, column Opus 4.6"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "80.8",
            "where": "Bar chart image; group 'SWE Verified (Resolved)'; value label on Claude-Opus-4.6-Max bar"
          },
          {
            "url": "https://www.minimax.io/news/minimax-m25",
            "name": "MiniMax M2.5",
            "title": "MiniMax M2.5: Built for Real-World Productivity.",
            "lab": "minimax",
            "date": "2026-02-12",
            "printed": "80.8",
            "where": "Overview bar chart image (8 panels); panel 'SWE-Bench Verified'; value label on Claude Opus 4.6 bar"
          },
          {
            "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
            "name": "Gemini 3.1 Pro",
            "title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
            "lab": "google",
            "date": "2026-02-19",
            "printed": "80.8%",
            "where": "Benchmark table image; row 'SWE-Bench Verified, Single attempt'; column claude-opus-4-6"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "80.8",
            "where": "Benchmark Table (appendix), row SWE-Bench Verified ; column Claude Opus 4.6 (max effort)"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
      },
      {
        "id": "claude-opus-4-6:hle:no-tools: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": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 40.0,
        "unit": "%",
        "printed": "40.0%",
        "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 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Opus 4.6', 'without 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-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "40.0%",
            "where": "Table 8.1.A (p.191), row Humanity's Last Exam / No tools, column Claude Opus 4.6"
          },
          {
            "url": "https://www.anthropic.com/news/claude-opus-4-7",
            "name": "Claude Opus 4.7",
            "title": "Introducing Claude Opus 4.7",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "40.0%",
            "where": "benchmark table image, row Humanity's Last Exam (no tools), column Opus 4.6"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "40.0",
            "where": "Bar chart image; group 'HLE (Pass@1), under 'Knowledge & Reasoning''; value label on Claude-Opus-4.6-Max bar"
          },
          {
            "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
            "name": "Gemini 3.1 Pro",
            "title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
            "lab": "google",
            "date": "2026-02-19",
            "printed": "40.0%",
            "where": "Benchmark table image; row 'Humanity's Last Exam, No tools'; column claude-opus-4-6"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "40.0",
            "where": "Benchmark Table (appendix), row HLE-Full (Reasoning & Knowledge; no tools) ; column Claude Opus 4.6 (max effort)"
          },
          {
            "url": "https://qwen.ai/blog?id=qwen3.7",
            "name": "Qwen3.7-Max",
            "title": "Qwen3.7: The Agent Frontier",
            "lab": "alibaba",
            "date": "2026-05-20",
            "printed": "40.0",
            "where": "Score chart image (Qwen3.7-Max-Score.png), panel HLE (Humanity's Last Exam; the post's table gives tool use as a separate row), value label on the Claude Opus-4.6 Max bar"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
      },
      {
        "id": "claude-opus-4-6:arc-agi-2:default: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": "arc-agi-2",
        "benchmark_name": "ARC-AGI-2",
        "setting": "default",
        "setting_name": "",
        "score": 68.8,
        "unit": "%",
        "printed": "68.8%",
        "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 'Novel problem-solving, ARC AGI 2', column 'Opus 4.6'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "max effort, 120K thinking budget",
        "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-4-6-system-card",
            "name": "Claude Opus 4.6 card",
            "title": "Claude Opus 4.6 System Card",
            "lab": "anthropic",
            "date": "2026-02-05",
            "printed": "68.8%",
            "where": "Table 2.3.A (p.18), row ARC-AGI-2 (Verified), column Claude Opus 4.6"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "68.8%",
            "where": "Table 8.1.A (p.192), row ARC-AGI-2, column Claude Opus 4.6"
          },
          {
            "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
            "name": "Gemini 3.1 Pro",
            "title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
            "lab": "google",
            "date": "2026-02-19",
            "printed": "68.8%",
            "where": "Benchmark table image; row 'ARC-AGI-2, ARC Prize Verified'; column claude-opus-4-6"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
      },
      {
        "id": "claude-opus-4-6:arc-agi-2:default: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": "arc-agi-2",
        "benchmark_name": "ARC-AGI-2",
        "setting": "default",
        "setting_name": "",
        "score": 69.17,
        "unit": "%",
        "printed": "69.17%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Claude Opus 4.6",
        "source": "https://www.anthropic.com/claude-opus-4-6-system-card",
        "source_title": "Claude Opus 4.6 System Card",
        "source_date": "2026-02-05",
        "where": "section 2.9 ARC-AGI, text",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "ARC Prize Foundation run: high effort, 120K thinking tokens, private set",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-16",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "claude-opus-4-6:gpqa-diamond:default: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": "gpqa-diamond",
        "benchmark_name": "GPQA Diamond",
        "setting": "default",
        "setting_name": "",
        "score": 91.3,
        "unit": "%",
        "printed": "91.3%",
        "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 'Graduate-level reasoning, GPQA Diamond', column 'Opus 4.6'",
        "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-4-6-system-card",
            "name": "Claude Opus 4.6 card",
            "title": "Claude Opus 4.6 System Card",
            "lab": "anthropic",
            "date": "2026-02-05",
            "printed": "91.3%",
            "where": "Table 2.3.A (p.18), row GPQA Diamond, column Claude Opus 4.6"
          },
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "91.3%",
            "where": "Table 8.1.A (p.191), row GPQA Diamond, column Claude Opus 4.6"
          },
          {
            "url": "https://www.anthropic.com/news/claude-opus-4-7",
            "name": "Claude Opus 4.7",
            "title": "Introducing Claude Opus 4.7",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "91.3%",
            "where": "benchmark table image, row GPQA Diamond, column Opus 4.6"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "91.3",
            "where": "Full benchmark table image; row 'GPQA Diamond (Pass@1)'; column 'Opus-4.6 Max'"
          },
          {
            "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
            "name": "Gemini 3.1 Pro",
            "title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
            "lab": "google",
            "date": "2026-02-19",
            "printed": "91.3%",
            "where": "Benchmark table image; row 'GPQA Diamond, No tools'; column claude-opus-4-6"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "91.3",
            "where": "Benchmark Table (appendix), row GPQA-Diamond ; column Claude Opus 4.6 (max effort)"
          },
          {
            "url": "https://z.ai/blog/glm-5.1",
            "name": "GLM-5.1",
            "title": "GLM-5.1: Towards Long-Horizon Tasks",
            "lab": "zai",
            "date": "2026-04-07",
            "printed": "91.3",
            "where": "Benchmark table, row 'GPQA-Diamond', column 'Claude Opus 4.6'"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
      },
      {
        "id": "claude-opus-4-6:hle:with-tools: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": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 53.0,
        "unit": "%",
        "printed": "53.0%",
        "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 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Opus 4.6', '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-4-6-system-card",
            "name": "Claude Opus 4.6 card",
            "title": "Claude Opus 4.6 System Card",
            "lab": "anthropic",
            "date": "2026-02-05",
            "printed": "53.0%",
            "where": "Figure 2.21.2.A (p.42), right panel With Tools, bar Opus 4.6 value label (web search/fetch, PTC, compaction up to 3M, max effort)"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "53.0",
            "where": "Benchmark Table (appendix), row HLE-Full w/ tools ; column Claude Opus 4.6 (max effort)"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
      },
      {
        "id": "claude-opus-4-6:hle:with-tools: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": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 53.3,
        "unit": "%",
        "printed": "53.3%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 4.6",
        "source": "https://www.anthropic.com/news/claude-opus-4-7",
        "source_title": "Introducing Claude Opus 4.7",
        "source_date": "2026-04-16",
        "where": "benchmark table (image), row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Opus 4.6', '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-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "53.3%",
            "where": "Table 8.1.A (p.191), row Humanity's Last Exam / With tools, column Claude Opus 4.6"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/527cd462f89b6de4.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/d434d15757c6abac1122af483617741776d5a114-2600x2638.png"
      },
      {
        "id": "claude-opus-4-6:browsecomp:default: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": "browsecomp",
        "benchmark_name": "BrowseComp",
        "setting": "default",
        "setting_name": "",
        "score": 84.0,
        "unit": "%",
        "printed": "84.0%",
        "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 'Agentic search, BrowseComp', column 'Opus 4.6'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "web search, web fetch, programmatic tool calling; compaction at 50K up to 10M tokens; max effort, no thinking",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.minimax.io/news/minimax-m25",
            "name": "MiniMax M2.5",
            "title": "MiniMax M2.5: Built for Real-World Productivity.",
            "lab": "minimax",
            "date": "2026-02-12",
            "printed": "84",
            "where": "Overview bar chart image (8 panels); panel 'BrowseComp (w/ctx)'; value label on Claude Opus 4.6 bar"
          },
          {
            "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
            "name": "Gemini 3.1 Pro",
            "title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
            "lab": "google",
            "date": "2026-02-19",
            "printed": "84.0%",
            "where": "Benchmark table image; row 'BrowseComp, Search + Python + Browse'; column claude-opus-4-6"
          },
          {
            "url": "https://z.ai/blog/glm-5.1",
            "name": "GLM-5.1",
            "title": "GLM-5.1: Towards Long-Horizon Tasks",
            "lab": "zai",
            "date": "2026-04-07",
            "printed": "84.0",
            "where": "Benchmark table, row 'BrowseComp w/ Context Manage', column 'Claude Opus 4.6'"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
      },
      {
        "id": "claude-opus-4-6:browsecomp:default: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": "browsecomp",
        "benchmark_name": "BrowseComp",
        "setting": "default",
        "setting_name": "",
        "score": 83.7,
        "unit": "%",
        "printed": "83.7%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 4.6",
        "source": "https://www.anthropic.com/claude-opus-4-6-system-card",
        "source_title": "Claude Opus 4.6 System Card",
        "source_date": "2026-02-05",
        "where": "Section 2.21.1, Figure 2.21.1.A 'BrowseComp', bar 'Opus 4.6', value label (page 39)",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "card update of 6 Mar 2026 after improved cheating detection; no thinking, max effort",
        "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-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "83.7%",
            "where": "Table 8.1.A (p.191), row BrowseComp, column Claude Opus 4.6"
          },
          {
            "url": "https://www.anthropic.com/news/claude-opus-4-7",
            "name": "Claude Opus 4.7",
            "title": "Introducing Claude Opus 4.7",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "83.7%",
            "where": "benchmark table image, row BrowseComp, column Opus 4.6"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "83.7",
            "where": "Full benchmark table image; row 'BrowseComp (Pass@1)'; column 'Opus-4.6 Max'"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "83.7",
            "where": "Benchmark Table (appendix), row BrowseComp ; column Claude Opus 4.6 (max effort)"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/a50a2643bd540b2c.png"
      },
      {
        "id": "claude-opus-4-6:tau2-bench:telecom: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": "tau2-bench",
        "benchmark_name": "τ²-bench",
        "setting": "telecom",
        "setting_name": "Telecom",
        "score": 99.3,
        "unit": "%",
        "printed": "99.3%",
        "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 'Agentic tool use, τ2-bench', Telecom line, column 'Opus 4.6'",
        "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-4-6-system-card",
            "name": "Claude Opus 4.6 card",
            "title": "Claude Opus 4.6 System Card",
            "lab": "anthropic",
            "date": "2026-02-05",
            "printed": "99.3%",
            "where": "Table 2.3.A (p.18), row τ²-bench (Telecom), column Claude Opus 4.6"
          },
          {
            "url": "https://www.minimax.io/news/minimax-m25",
            "name": "MiniMax M2.5",
            "title": "MiniMax M2.5: Built for Real-World Productivity.",
            "lab": "minimax",
            "date": "2026-02-12",
            "printed": "99.3",
            "where": "Search and tool-calling bar chart image; panel 'τ² Telecom'; value label on Claude Opus 4.6 bar"
          },
          {
            "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
            "name": "Gemini 3.1 Pro",
            "title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
            "lab": "google",
            "date": "2026-02-19",
            "printed": "99.3%",
            "where": "Benchmark table image; row 'τ2-bench, Telecom'; column claude-opus-4-6"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
      },
      {
        "id": "claude-opus-4-6:swe-bench-pro:default: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": "swe-bench-pro",
        "benchmark_name": "SWE-bench Pro",
        "setting": "default",
        "setting_name": "",
        "score": 53.4,
        "unit": "%",
        "printed": "53.4%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Opus 4.6",
        "source": "https://www.anthropic.com/news/claude-opus-4-7",
        "source_title": "Introducing Claude Opus 4.7",
        "source_date": "2026-04-16",
        "where": "benchmark table (image), row 'Agentic coding, SWE-bench Pro', column 'Opus 4.6'",
        "own": false,
        "primary": true,
        "parallel": false,
        "notes": "memorization screens flag a subset of problems",
        "caveat": "The source doesn't say whether this is the public or the commercial set",
        "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-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "53.4%",
            "where": "section 8.1 Evaluation summary table, SWE-bench Pro row, Claude Opus 4.6 column"
          }
        ],
        "source_image": "https://frontierlog.ai/sources/527cd462f89b6de4.png",
        "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/d434d15757c6abac1122af483617741776d5a114-2600x2638.png"
      }
    ]
  }
}