{
  "api_version": "1",
  "updated": "2026-10-17",
  "latest_event": "2026-10-15",
  "rights": "© Frontier Log, all rights reserved.",
  "benchmark": {
    "slug": "browsecomp",
    "name": "BrowseComp",
    "family": "browsecomp",
    "unit": "%",
    "higher_is_better": true,
    "status": "active",
    "benchmark_url": "https://openai.com/index/browsecomp/",
    "settings": [
      {
        "setting": "default",
        "name": "",
        "url": "https://frontierlog.ai/benchmark/browsecomp",
        "results": 30
      }
    ]
  },
  "total": 30,
  "results": [
    {
      "id": "claude-sonnet-4-5:browsecomp:default:1",
      "model": "claude-sonnet-4-5",
      "model_name": "Claude Sonnet 4.5",
      "model_date": "2025-09-29",
      "event": "https://frontierlog.ai/event/2025-09-29-claude-sonnet-4-5",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 43.9,
      "unit": "%",
      "printed": "43.9%",
      "reported_by": "anthropic",
      "label": "Sonnet 4.5",
      "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 'Sonnet 4.5'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "Sonnet 4.5's own post and system card print no BrowseComp; the Opus 4.6 system card (Figure 2.21.1.A) prints the same value",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
      "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
    },
    {
      "id": "deepseek-v3-2-exp:browsecomp:default:1",
      "model": "deepseek-v3-2-exp",
      "model_name": "DeepSeek-V3.2-Exp",
      "model_date": "2025-09-29",
      "event": "https://frontierlog.ai/event/2025-09-29-deepseek-v3-2-exp",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 40.1,
      "unit": "%",
      "printed": "40.1",
      "reported_by": "deepseek",
      "label": "DeepSeek-V3.2-Exp",
      "source": "https://api-docs.deepseek.com/news/news250929/",
      "source_title": "Introducing DeepSeek-V3.2-Exp",
      "source_date": "2025-09-29",
      "where": "benchmark table image, row 'BrowseComp' (Search Agent), column 'DeepSeek-V3.2-Exp'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/4d9152a1f18ce8ee.webp",
      "source_image_original": "https://api-docs.deepseek.com/img/v3_2_benchmark.webp"
    },
    {
      "id": "kimi-k2-thinking:browsecomp:default:1",
      "model": "kimi-k2-thinking",
      "model_name": "Kimi K2 Thinking",
      "model_date": "2025-11-06",
      "event": "https://frontierlog.ai/event/2025-11-06-kimi-k2-thinking",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 60.2,
      "unit": "%",
      "printed": "60.2",
      "reported_by": "moonshot",
      "label": "K2 Thinking",
      "source": "https://www.kimi.ai/blog/kimi-k2-thinking",
      "source_title": "Introducing Kimi K2 Thinking",
      "source_date": "2025-11-06",
      "where": "'Full Evaluations' table, row 'BrowseComp' (Agentic Search Tasks), 'w/ tools', column 'K2 Thinking'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "search, code-interpreter and web-browsing tools; up to 300 steps, 24k reasoning tokens per step; context management hides all previous tool outputs when input exceeds 256k (footnote 4e); Hugging Face blocked",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "gemini-3-pro:browsecomp:default:1",
      "model": "gemini-3-pro",
      "model_name": "Gemini 3 Pro",
      "model_date": "2025-11-18",
      "event": "https://frontierlog.ai/event/2025-11-18-gemini-3-pro",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 59.2,
      "unit": "%",
      "printed": "59.2%",
      "reported_by": "google",
      "label": "Gemini 3 Pro",
      "source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
      "source_title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
      "source_date": "2026-02-19",
      "where": "benchmark table (image), row 'BrowseComp' (Search + Python + Browse), column 'Gemini 3 Pro Thinking (High)'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "Gemini 3 Pro's own post prints no BrowseComp; earliest Google report. Anthropic's Opus 4.6 table labels the same 59.2% as Gemini 3 Pro '(Deep Research)'",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/6b612deebae92383.gif",
      "source_image_original": "https://storage.googleapis.com/gweb-uniblog-publish-prod/original_images/gemini_3-1-pro__benchmarks.gif"
    },
    {
      "id": "claude-opus-4-5:browsecomp:default: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": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 67.8,
      "unit": "%",
      "printed": "67.8%",
      "reported_by": "anthropic",
      "label": "Opus 4.5",
      "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 4.5'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "Opus 4.5's own post and card report only BrowseComp-Plus (a different benchmark); the Opus 4.6 system card (Figure 2.21.1.A) prints the same value",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/e6c54904fe1e4f51.png",
      "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/f9564dd2f758237bd9dbe775674c4a375aff1e8a-2600x2968.png"
    },
    {
      "id": "gpt-5-2:browsecomp:default: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": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 65.8,
      "unit": "%",
      "printed": "65.8%",
      "reported_by": "openai",
      "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": "agentic tool-use table (columns GPT-5.2 Thinking / GPT-5.2 Pro / GPT-5.1 Thinking), row 'BrowseComp', column 'GPT-5.2 Thinking'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Pro and Ultra variants are separate columns, not recorded",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "kimi-k2-5:browsecomp:default:1",
      "model": "kimi-k2-5",
      "model_name": "Kimi K2.5",
      "model_date": "2026-01-27",
      "event": "https://frontierlog.ai/event/2026-01-27-kimi-k2-5",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 74.9,
      "unit": "%",
      "printed": "74.9",
      "reported_by": "moonshot",
      "label": "Kimi K2.5 (Thinking)",
      "source": "https://www.kimi.ai/blog/kimi-k2-5",
      "source_title": "Kimi K2.5: Visual Agentic Intelligence",
      "source_date": "2026-01-27",
      "where": "Appendix 'Benchmark table', row 'BrowseComp (w/ctx mgm)' (Agentic Search), column 'Kimi K2.5 (Thinking)'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "with context management, discard-all strategy (footnote 3); the headline chart at the top of the post shows this figure",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "kimi-k2-5:browsecomp:default:2",
      "model": "kimi-k2-5",
      "model_name": "Kimi K2.5",
      "model_date": "2026-01-27",
      "event": "https://frontierlog.ai/event/2026-01-27-kimi-k2-5",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 60.6,
      "unit": "%",
      "printed": "60.6",
      "reported_by": "moonshot",
      "label": "Kimi K2.5 (Thinking)",
      "source": "https://www.kimi.ai/blog/kimi-k2-5",
      "source_title": "Kimi K2.5: Visual Agentic Intelligence",
      "source_date": "2026-01-27",
      "where": "Appendix 'Benchmark table', first row 'BrowseComp' (no qualifier, Agentic Search), column 'Kimi K2.5 (Thinking)'",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "no context management: tasks exceeding the context length counted as failed",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "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",
      "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, context compaction triggered at 50k tokens up to 10M total tokens, max effort, no thinking (footnote)",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "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",
      "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 updated March 6, 2026 after an improved cheating-detection pipeline (changelog: 83.97% -> 83.73%); thinking disabled, compaction at 50k up to 10M tokens, max effort",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "minimax-m2-5:browsecomp:default:1",
      "model": "minimax-m2-5",
      "model_name": "MiniMax M2.5",
      "model_date": "2026-02-12",
      "event": "https://frontierlog.ai/event/2026-02-12-minimax-m2-5",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 76.3,
      "unit": "%",
      "printed": "76.3%",
      "reported_by": "minimax",
      "label": "MiniMax M2.5",
      "source": "https://www.minimax.io/news/minimax-m25",
      "source_title": "MiniMax M2.5: Built for Real-World Productivity.",
      "source_date": "2026-02-12",
      "where": "opening paragraph, list of scores ('… in BrowseComp (with context management)')",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "with context management: WebExplorer agent framework; when token usage exceeds 30% of the maximum context, all history is discarded (evaluation notes)",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "gemini-3-1-pro:browsecomp:default:1",
      "model": "gemini-3-1-pro",
      "model_name": "Gemini 3.1 Pro",
      "model_date": "2026-02-19",
      "event": "https://frontierlog.ai/event/2026-02-19-gemini-3-1-pro",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 85.9,
      "unit": "%",
      "printed": "85.9%",
      "reported_by": "google",
      "label": "Gemini 3.1 Pro",
      "source": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
      "source_title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
      "source_date": "2026-02-19",
      "where": "benchmark table (image), row 'BrowseComp' (Search + Python + Browse), column 'Gemini 3.1 Pro Thinking (High)'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Thinking (High); search + Python + browse",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/6b612deebae92383.gif",
      "source_image_original": "https://storage.googleapis.com/gweb-uniblog-publish-prod/original_images/gemini_3-1-pro__benchmarks.gif"
    },
    {
      "id": "gpt-5-4:browsecomp:default: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": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 82.7,
      "unit": "%",
      "printed": "82.7%",
      "reported_by": "openai",
      "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 (columns GPT‑5.4 / GPT‑5.4Pro / GPT‑5.3-Codex / GPT‑5.2 / GPT‑5.2Pro), row 'BrowseComp', column 'GPT‑5.4'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Pro and Ultra variants are separate columns, not recorded",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "claude-mythos-preview:browsecomp:default:1",
      "model": "claude-mythos-preview",
      "model_name": "Claude Mythos Preview",
      "model_date": "2026-04-07",
      "event": "https://frontierlog.ai/event/2026-04-07-claude-mythos-preview",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 86.9,
      "unit": "%",
      "printed": "86.9%",
      "reported_by": "anthropic",
      "label": "Mythos Preview",
      "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 search, BrowseComp', column 'Mythos Preview'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "the Mythos Preview post links no system card and prints no BrowseComp; Opus 4.7 post is the earliest report",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/527cd462f89b6de4.png",
      "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/d434d15757c6abac1122af483617741776d5a114-2600x2638.png"
    },
    {
      "id": "claude-mythos-preview:browsecomp:default:2",
      "model": "claude-mythos-preview",
      "model_name": "Claude Mythos Preview",
      "model_date": "2026-04-07",
      "event": "https://frontierlog.ai/event/2026-04-07-claude-mythos-preview",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 87.9,
      "unit": "%",
      "printed": "87.9",
      "reported_by": "anthropic",
      "label": "Mythos Preview",
      "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 8.1 evaluation summary table (page 251), row 'BrowseComp', column 'Mythos Preview'",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "claude-opus-4-7:browsecomp:default: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": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 79.3,
      "unit": "%",
      "printed": "79.3%",
      "reported_by": "anthropic",
      "label": "Opus 4.7",
      "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 search, BrowseComp', column 'Opus 4.7'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "system card 8.8.2: thinking off, max effort, 10M token limit, context compaction triggered at 200k tokens",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "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-4-7:browsecomp:default: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": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 79.8,
      "unit": "%",
      "printed": "79.8",
      "reported_by": "anthropic",
      "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": "Section 8.1 evaluation summary table (page 194), row 'BrowseComp', column 'Claude Opus 4.7'",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "re-run: footnote 27 'Changes to the Opus 4.7 BrowseComp score are due to a new blocklist, context compaction at 200k tokens and using adaptive thinking.'",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "kimi-k2-6:browsecomp:default:1",
      "model": "kimi-k2-6",
      "model_name": "Kimi K2.6",
      "model_date": "2026-04-20",
      "event": "https://frontierlog.ai/event/2026-04-20-kimi-k2-6",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 83.2,
      "unit": "%",
      "printed": "83.2",
      "reported_by": "moonshot",
      "label": "Kimi K2.6",
      "source": "https://www.kimi.ai/blog/kimi-k2-6",
      "source_title": "Kimi K2.6: Advancing Open-Source Coding",
      "source_date": "2026-04-20",
      "where": "'Benchmark Table' near the end, row 'BrowseComp' (Agentic), column 'Kimi K2.6'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "context management, discard-all strategy as for Kimi K2.5 and DeepSeek-V3.2 (footnote 3)",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "gpt-5-5:browsecomp:default: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": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 84.4,
      "unit": "%",
      "printed": "84.4%",
      "reported_by": "openai",
      "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 (columns GPT‑5.5 / GPT‑5.4 / GPT‑5.5 Pro / GPT‑5.4 Pro / Claude Opus 4.7 / Gemini 3.1 Pro), row 'BrowseComp', column 'GPT‑5.5'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Pro and Ultra variants are separate columns, not recorded",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "deepseek-v4-pro:browsecomp:default:1",
      "model": "deepseek-v4-pro",
      "model_name": "DeepSeek-V4-Pro",
      "model_date": "2026-04-24",
      "event": "https://frontierlog.ai/event/2026-04-24-deepseek-v4-preview",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 83.4,
      "unit": "%",
      "printed": "83.4",
      "reported_by": "deepseek",
      "label": "DS-V4-Pro Max",
      "source": "https://api-docs.deepseek.com/news/news260424/",
      "source_title": "DeepSeek V4 Preview Release",
      "source_date": "2026-04-24",
      "where": "benchmark table image, row 'BrowseComp (Pass@1)' (Agentic), column 'DS-V4-Pro Max'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Max reasoning effort; context management not stated",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/9226203de8e24b33.png",
      "source_image_original": "https://api-docs.deepseek.com/img/v4-benchmark-2.png"
    },
    {
      "id": "claude-opus-4-8:browsecomp:default: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": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 84.3,
      "unit": "%",
      "printed": "84.3",
      "reported_by": "anthropic",
      "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": "Section 8.1 evaluation summary table (page 194), row 'BrowseComp', column 'Claude Opus 4.8', the '(single-agent)' value",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "single-agent; adaptive thinking, max effort, 10M-token limit, context compaction triggered at 200k tokens (8.10.2). The Opus 4.8 post's table has no BrowseComp row",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "minimax-m3:browsecomp:default:1",
      "model": "minimax-m3",
      "model_name": "MiniMax M3",
      "model_date": "2026-06-01",
      "event": "https://frontierlog.ai/event/2026-06-01-minimax-m3",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 83.52,
      "unit": "%",
      "printed": "83.52",
      "reported_by": "minimax",
      "label": "MiniMax M3",
      "source": "https://www.minimax.io/blog/minimax-m3",
      "source_title": "MiniMax M3: Frontier Coding, 1M Context, Native Multimodality — All in One Model",
      "source_date": "2026-06-01",
      "where": "benchmark table image, row 'BrowseComp' (Cowork (Agent)), column 'MiniMax M3'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "WebExplorer agent framework; when token usage exceeds 64K, all history is discarded (evaluation notes)",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/5720f3ad06ea2ce9.jpg",
      "source_image_original": "https://filecdn.minimax.chat/public/img_v3_02128_b7726cd8-879a-4b7a-a9da-db4395ea597g-1780272508686.jpg"
    },
    {
      "id": "claude-fable-5:browsecomp:default: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": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 87.4,
      "unit": "%",
      "printed": "87.4%",
      "reported_by": "anthropic",
      "label": "Fable 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 'Fable 5'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "Fable 5's own post has no BrowseComp row and its card prints '-' for Fable 5; the Opus 5 card's summary table prints the same value",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
      "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
    },
    {
      "id": "claude-mythos-5:browsecomp:default:1",
      "model": "claude-mythos-5",
      "model_name": "Claude Mythos 5",
      "model_date": "2026-06-09",
      "event": "https://frontierlog.ai/event/2026-06-09-claude-fable-5-mythos-5",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 88.0,
      "unit": "%",
      "printed": "88.0",
      "reported_by": "anthropic",
      "label": "Mythos 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 8.1 evaluation summary table (page 251), row 'BrowseComp', column 'Mythos 5', the '(single-agent)' value",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "single-agent; adaptive thinking, max effort, 10M-token limit, compaction at 200k tokens (8.14.2). The launch post's table has no BrowseComp row",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "claude-sonnet-5:browsecomp:default:1",
      "model": "claude-sonnet-5",
      "model_name": "Claude Sonnet 5",
      "model_date": "2026-06-30",
      "event": "https://frontierlog.ai/event/2026-06-30-claude-sonnet-5",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 84.7,
      "unit": "%",
      "printed": "84.7",
      "reported_by": "anthropic",
      "label": "Claude Sonnet 5",
      "source": "https://www.anthropic.com/claude-sonnet-5-system-card",
      "source_title": "Claude Sonnet 5 System Card",
      "source_date": "2026-06-30",
      "where": "Section 8.1 evaluation summary table (page 115), row 'BrowseComp', column 'Claude Sonnet 5', the '(single agent)' value",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "adaptive thinking, max effort, 10M-token limit with context compaction triggered at 200k (Table 8.1.A caption). The post shows only an effort-level curve without values",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "gpt-5-6-sol:browsecomp:default:1",
      "model": "gpt-5-6-sol",
      "model_name": "GPT-5.6 Sol",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 90.4,
      "unit": "%",
      "printed": "90.4%",
      "reported_by": "openai",
      "label": "GPT‑5.6 Sol",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "eval table with columns GPT‑5.6 Sol / Sol Ultra / Terra / Luna / …, row 'BrowseComp', column 'GPT‑5.6 Sol'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Pro and Ultra variants are separate columns, not recorded",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "kimi-k3:browsecomp:default:1",
      "model": "kimi-k3",
      "model_name": "Kimi K3",
      "model_date": "2026-07-16",
      "event": "https://frontierlog.ai/event/2026-07-16-kimi-k3",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 91.2,
      "unit": "%",
      "printed": "91.2",
      "reported_by": "moonshot",
      "label": "Kimi K3 (max)",
      "source": "https://www.kimi.ai/blog/kimi-k3",
      "source_title": "Kimi K3: Open Frontier Intelligence",
      "source_date": "2026-07-16",
      "where": "'Full Benchmark Table', row 'BrowseComp', column 'Kimi K3 (max)'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "context compaction as in the Claude model cards, triggered at 300K tokens (footnote 'BrowseComp'); reasoning effort max",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "id": "kimi-k3:browsecomp:default:2",
      "model": "kimi-k3",
      "model_name": "Kimi K3",
      "model_date": "2026-07-16",
      "event": "https://frontierlog.ai/event/2026-07-16-kimi-k3",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 90.4,
      "unit": "%",
      "printed": "90.4",
      "reported_by": "moonshot",
      "label": "Kimi K3",
      "source": "https://www.kimi.ai/blog/kimi-k3",
      "source_title": "Kimi K3: Open Frontier Intelligence",
      "source_date": "2026-07-16",
      "where": "Footnotes, 'Productivity and agentic benchmarks', the 'BrowseComp.' note (score with a 1M-token context window and no context management)",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "1M-token context window, no context management",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    },
    {
      "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",
      "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": "context compaction triggered at 200k tokens (card 8.10.2); the card's summary table prints the same value",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "source_image": "https://frontierlog.ai/sources/924165396ca89130.png",
      "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a8fb4f77a9fe240e6f27f3bdc47a137f3c74a29d-2600x2578.png"
    },
    {
      "id": "gpt-6-astra:browsecomp:default:1",
      "model": "gpt-6-astra",
      "model_name": "GPT-6 Astra",
      "model_date": "2026-09-03",
      "event": "https://frontierlog.ai/event/2026-09-03-gpt-6-astra",
      "benchmark": "browsecomp",
      "benchmark_name": "BrowseComp",
      "setting": "default",
      "setting_name": "",
      "score": 91.5,
      "unit": "%",
      "printed": "91.5%",
      "reported_by": "openai",
      "label": "GPT‑6 Astra",
      "source": "https://openai.com/index/gpt-6-astra",
      "source_title": "GPT-6 Astra: A new generation of intelligence",
      "source_date": "2026-09-03",
      "where": "'Professional' eval table, row 'BrowseComp', column 'GPT‑6 Astra'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "Pro and Ultra variants are separate columns, not recorded",
      "human_verified": true,
      "human_verified_on": "2026-10-17"
    }
  ]
}