{ "api_version": "1", "updated": "2026-11-10", "latest_event": "2026-11-09", "rights": "© Frontier Log, all rights reserved.", "compare": { "slug": "claude-opus-5-5-vs-gpt-6-astra", "models": [ { "slug": "claude-opus-5-5", "name": "Claude Opus 5.5", "lab": "anthropic", "date": "2026-09-22", "source": "https://www.anthropic.com/claude-opus-5-5", "source_title": "Introducing Claude Opus 5.5", "flagship": true, "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "releases": [], "variant_of": null, "made_by": "Anthropic", "url": "https://frontierlog.ai/model/claude-opus-5-5", "results": 18, "price": [ { "from": "2026-09-22", "listed": false, "currency": "USD", "input_per_1m": 4.0, "cached_input_per_1m": 0.2, "cached_input_derived": false, "output_per_1m": 20.0, "cache": { "kind": "price" }, "tiers": [], "printed": { "input": "$4", "cached_input": "$0.20", "output": "$20" }, "printed_unit": "1M tokens", "notes": "", "sources": [ "https://www.anthropic.com/claude-opus-5-5", "https://www.anthropic.com/claude-sonnet-5-5", "https://platform.claude.com/docs/en/about-claude/pricing" ], "check": "corroborated" } ], "context": { "from": "2026-09-22", "listed": false, "window_tokens": 1000000, "input_tokens": null, "output_tokens": 128000, "tiers": [ { "condition": "Message Batches API with the 300k output beta", "window_tokens": null, "input_tokens": null, "output_tokens": 300000, "printed": { "window": null, "input": null, "output": "300K" } } ], "printed": { "window": "1M", "input": null, "output": "128k" }, "notes": "", "sources": [ "https://platform.claude.com/docs/en/release-notes/overview", "https://platform.claude.com/docs/en/models/opus-5-5/overview", "https://platform.claude.com/docs/en/about-claude/models/overview" ], "check": "corroborated" } }, { "slug": "gpt-6-astra", "name": "GPT-6 Astra", "lab": "openai", "date": "2026-09-03", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "flagship": true, "event": "https://frontierlog.ai/event/2026-09-03-gpt-6-astra", "releases": [], "variant_of": null, "made_by": "OpenAI", "url": "https://frontierlog.ai/model/gpt-6-astra", "results": 61, "price": [ { "from": "2026-09-03", "listed": false, "currency": "USD", "input_per_1m": 10.0, "cached_input_per_1m": 1.0, "cached_input_derived": false, "output_per_1m": 50.0, "cache": { "kind": "price" }, "tiers": [ { "over": 272000, "measure": "input", "input_per_1m": 20.0, "cached_input_per_1m": 2.0, "cached_input_derived": false, "output_per_1m": 75.0, "printed": { "input": "$20.00", "cached_input": "$2.00", "output": "$75.00" } } ], "printed": { "input": "$10", "cached_input": "$1.00", "output": "$50" }, "printed_unit": "1M tokens", "notes": "", "sources": [ "https://openai.com/index/gpt-6-astra", "https://openai.com/index/introducing-gpt-6-1-sol", "https://developers.openai.com/api/docs/pricing", "https://developers.openai.com/api/docs/models/gpt-6-astra.md", "https://developers.openai.com/api/docs/models", "https://developers.openai.com/api/docs/models/gpt-6-astra" ], "check": "corroborated" } ], "context": { "from": "2026-11-04", "listed": true, "window_tokens": 1050000, "input_tokens": 922000, "output_tokens": 128000, "tiers": [], "printed": { "window": "1,050,000", "input": "922,000", "output": "128,000" }, "notes": "", "sources": [ "https://developers.openai.com/api/docs/models/gpt-6-astra.md", "https://developers.openai.com/api/docs/models/compare?model=gpt-6-astra", "https://developers.openai.com/api/docs/models" ], "check": "corroborated" } } ], "url": "https://frontierlog.ai/compare-models/claude-opus-5-5-vs-gpt-6-astra", "api": "https://frontierlog.ai/api/v1/compare-models/claude-opus-5-5-vs-gpt-6-astra", "both": [ { "benchmark": "terminal-bench-4-0", "benchmark_name": "Terminal-Bench 4.0", "setting": "default", "setting_name": "", "unit": "%", "url": "https://frontierlog.ai/benchmark/terminal-bench-4-0", "results": [ { "id": "claude-opus-5-5:terminal-bench-4-0:default:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "terminal-bench-4-0", "benchmark_name": "Terminal-Bench 4.0", "setting": "default", "setting_name": "", "score": 66.4, "unit": "%", "printed": "66.4%", "reported_by": "anthropic", "run_by": null, "label": "Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5", "source_title": "Introducing Claude Opus 5.5", "source_date": "2026-09-22", "where": "benchmark table, row 'Terminal-Bench 4.0', column 'Opus 5.5'", "own": true, "primary": true, "parallel": false, "notes": "xhigh effort; standard error ±2.6 points", "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-haiku-5-5-system-card", "name": "Claude Haiku 5.5 card", "title": "Claude Haiku 5.5 System Card", "lab": "anthropic", "date": "2026-10-07", "printed": "66.4%", "where": "Section 8.4 (p.115-116) sentence: \"Claude Opus 5.5 scored 66.4%\" (xhigh effort)" }, { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "66.4", "where": "Table 8.1.A (p.174), row Terminal-Bench 4.0, column Claude Opus 5.5 (caption: reported at xhigh effort)" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "66.4%", "where": "Section 8.5 (p.113) sentence: \"Claude Opus 5.5 scored 66.4%\" (xhigh effort)" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5", "name": "Claude Sonnet 5.5", "title": "Introducing Claude Sonnet 5.5", "lab": "anthropic", "date": "2026-09-28", "printed": "66.4%", "where": "HTML benchmark table under 'Performance', row Terminal-Bench 4.0, column Opus 5.5 (footnote 1: Xhigh effort)" }, { "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/", "name": "Gemini 4 Argon", "title": "Gemini 4 Argon: our next era of frontier intelligence", "lab": "google", "date": "2026-09-30", "printed": "66.4%", "where": "Benchmark table image; row 'Terminal-bench 4.0'; column claude-opus-5-5" } ], "source_capture": "https://frontierlog.ai/captures/ca3ad55d7ec61f3c.png" }, { "id": "gpt-6-astra:terminal-bench-4-0: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": "terminal-bench-4-0", "benchmark_name": "Terminal-Bench 4.0", "setting": "default", "setting_name": "", "score": 57.9, "unit": "%", "printed": "57.9%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "Coding table, row 'Terminal-Bench 4.0', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "high effort", "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-haiku-5-5-system-card", "name": "Claude Haiku 5.5 card", "title": "Claude Haiku 5.5 System Card", "lab": "anthropic", "date": "2026-10-07", "printed": "57.9%", "where": "Section 8.4 (p.115-116) sentence: \"OpenAI reported GPT-6 Astra's final Terminal-Bench 4.0 score at 57.9%, scored at high thinking effort\"" }, { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "57.9", "where": "Table 8.1.A (p.174), row Terminal-Bench 4.0, column GPT-6 Astra" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "57.9%", "where": "Section 8.5 (p.113) sentence: \"GPT-6 Astra reported their final Terminal-Bench 4.0 score at 57.9%, scored at high thinking effort\"" }, { "url": "https://www.anthropic.com/claude-opus-5-5", "name": "Claude Opus 5.5", "title": "Introducing Claude Opus 5.5", "lab": "anthropic", "date": "2026-09-22", "printed": "57.9%", "where": "HTML benchmark table under 'Performance and cost-effectiveness', row Terminal-Bench 4.0¹, column GPT-6 Astra (high effort, as reported by OpenAI)" }, { "url": "https://openai.com/index/introducing-gpt-6-1-terra", "name": "GPT-6.1 Terra", "title": "Introducing GPT-6.1 Terra", "lab": "openai", "date": "2026-10-26", "printed": "57.9%", "where": "\"Terminal-Bench 4.0\" chart (score vs cost per task; no value labels): point 'GPT-6 Astra', effort 'Xhigh', read from the point's aria-label (load the page in a browser and hover or inspect the SVG point)" } ], "source_capture": "https://frontierlog.ai/captures/4de8f321fbadf8b8.png" } ] }, { "benchmark": "terminal-bench-science-0-1", "benchmark_name": "Terminal-Bench Science 0.1", "setting": "default", "setting_name": "", "unit": "%", "url": "https://frontierlog.ai/benchmark/terminal-bench-science-0-1", "results": [ { "id": "claude-opus-5-5:terminal-bench-science-0-1:default:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "terminal-bench-science-0-1", "benchmark_name": "Terminal-Bench Science 0.1", "setting": "default", "setting_name": "", "score": 58.7, "unit": "%", "printed": "58.7%", "reported_by": "anthropic", "run_by": null, "label": "Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5", "source_title": "Introducing Claude Opus 5.5", "source_date": "2026-09-22", "where": "benchmark table, row 'Terminal-Bench-Science 0.1', column 'Opus 5.5'", "own": true, "primary": true, "parallel": false, "notes": "max effort; standard error ±3.5 to 5 points", "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-haiku-5-5-system-card", "name": "Claude Haiku 5.5 card", "title": "Claude Haiku 5.5 System Card", "lab": "anthropic", "date": "2026-10-07", "printed": "58.7%", "where": "Section 8.5 (p.116) sentence: \"Claude Opus 5.5 scored 58.7%\"" }, { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "58.7", "where": "Table 8.1.A (p.174), row Terminal-Bench-Science 0.1, column Claude Opus 5.5" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "58.7%", "where": "Section 8.6 (p.114) sentence: \"Claude Opus 5.5 scored 58.7%\"" } ], "source_capture": "https://frontierlog.ai/captures/ca3ad55d7ec61f3c.png" }, { "id": "gpt-6-astra:terminal-bench-science-0-1: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": "terminal-bench-science-0-1", "benchmark_name": "Terminal-Bench Science 0.1", "setting": "default", "setting_name": "", "score": 64.6, "unit": "%", "printed": "64.6%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "Academic table, row 'Terminal-Bench Science 0.1', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "max effort", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-16", "check": "human", "corroborated": true, "corroborated_on": "2026-10-19", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "64.6", "where": "Table 8.1.A (p.174), row Terminal-Bench-Science 0.1, column GPT-6 Astra" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "64.6%", "where": "Section 8.6 (p.114) sentence: \"OpenAI reported GPT-6 Astra at 64.6% with max thinking effort\"" }, { "url": "https://www.anthropic.com/claude-opus-5-5", "name": "Claude Opus 5.5", "title": "Introducing Claude Opus 5.5", "lab": "anthropic", "date": "2026-09-22", "printed": "64.6%", "where": "HTML benchmark table under 'Performance and cost-effectiveness', row Terminal-Bench-Science 0.1³, column GPT-6 Astra (as reported by OpenAI)" } ], "source_capture": "https://frontierlog.ai/captures/478a8e3e7a393b2d.png" } ] }, { "benchmark": "automationbench-1-0-6", "benchmark_name": "AutomationBench 1.0.6", "setting": "private", "setting_name": "Private set", "unit": "%", "url": "https://frontierlog.ai/benchmark/automationbench-1-0-6", "results": [ { "id": "claude-opus-5-5:automationbench-1-0-6:private:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "automationbench-1-0-6", "benchmark_name": "AutomationBench 1.0.6", "setting": "private", "setting_name": "Private set", "score": 40.0, "unit": "%", "printed": "40.0%", "reported_by": "anthropic", "run_by": "Zapier", "label": "Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5", "source_title": "Introducing Claude Opus 5.5", "source_date": "2026-09-22", "where": "benchmark table, row 'AutomationBench', column 'Opus 5.5'", "own": true, "primary": true, "parallel": false, "notes": "without fallback models; set and release 1.0.6 named in the Claude Sonnet 5.5 system card, which prints the same figure", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-16", "check": "human", "corroborated": true, "corroborated_on": "2026-10-19", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "40.0", "where": "Table 8.1.A (p.174), row AutomationBench, column Claude Opus 5.5" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "40.0%", "where": "Section 8.14.6 (p.137) sentence: \"(40.0% without the re-run, as reported in the Claude Opus 5.5 System Card)\"" } ], "source_capture": "https://frontierlog.ai/captures/ca3ad55d7ec61f3c.png" }, { "id": "gpt-6-astra:automationbench-1-0-6:private: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": "automationbench-1-0-6", "benchmark_name": "AutomationBench 1.0.6", "setting": "private", "setting_name": "Private set", "score": 41.4, "unit": "%", "printed": "41.4%", "reported_by": "openai", "run_by": "Zapier", "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 table, row 'AutomationBench', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "highest score at any effort; set named in the GPT-6.1 Sol post, which prints the same figure", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-16", "check": "human", "corroborated": true, "corroborated_on": "2026-10-19", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "41.4", "where": "Table 8.1.A (p.174), row AutomationBench, column GPT-6 Astra" }, { "url": "https://www.anthropic.com/claude-opus-5-5", "name": "Claude Opus 5.5", "title": "Introducing Claude Opus 5.5", "lab": "anthropic", "date": "2026-09-22", "printed": "41.4%", "where": "HTML benchmark table under 'Performance and cost-effectiveness', row AutomationBench² (run and reported by Zapier), column GPT-6 Astra" }, { "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/", "name": "Gemini 4 Argon", "title": "Gemini 4 Argon: our next era of frontier intelligence", "lab": "google", "date": "2026-09-30", "printed": "41.4%", "where": "Benchmark table image; row 'AutomationBench, Score'; column gpt-6-astra" } ], "source_capture": "https://frontierlog.ai/captures/8618c81cd990eeed.png" } ] }, { "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/offline-partial", "results": [ { "id": "claude-opus-5-5:osworld-2-1:offline-partial:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-partial", "setting_name": "Offline set, partial score", "score": 87.2, "unit": "%", "printed": "87.2%", "reported_by": "anthropic", "run_by": null, "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-haiku-5-5-system-card", "source_title": "Claude Haiku 5.5 System Card", "source_date": "2026-10-07", "where": "section 8.9.3 OSWorld 2.1 offline subset, text (PDF page 128), the Claude Opus 5.5 partial score", "own": false, "primary": true, "parallel": false, "notes": "82-task offline subset, no internet access; max effort; mean of 5 attempts", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-20", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/78b2fca2e8e996c4.png" }, { "id": "gpt-6-astra:osworld-2-1:offline-partial:5", "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": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-partial", "setting_name": "Offline set, partial score", "score": 80.9, "unit": "%", "printed": "80.9%", "reported_by": "openai", "run_by": null, "label": "GPT-6 Astra (Max)", "source": "https://openai.com/index/introducing-gpt-6-1-terra", "source_title": "Introducing GPT-6.1 Terra", "source_date": "2026-10-26", "where": "\"OSWorld 2.1, offline set\" chart (score vs cost per task; no value labels): point 'GPT-6 Astra', effort 'Max', read from the point's aria-label (load the page in a browser and hover or inspect the SVG point)", "own": false, "primary": true, "parallel": false, "notes": "max effort", "caveat": "", "human_verified": false, "human_verified_on": null, "check": "corroborated", "corroborated": true, "corroborated_on": "2026-11-07", "corroborated_by": [ { "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/", "name": "Gemini 4 Carbon", "title": "Gemini 4 Carbon: our next frontier model", "lab": "google", "date": "2026-10-28", "printed": "80.9%", "where": "row \"OSWorld-2.1 (Offline subset)\", evals table image on the post (also page 5 of https://storage.googleapis.com/deepmind-media/gemini/gemini_4_carbon_model_evaluation.pdf), column GPT-6 Astra; methodology: non-Gemini models at maximum reasoning settings unless unavailable; Astra offline subset score \"sourced from their respective published evaluations\"" } ], "source_capture": "https://frontierlog.ai/captures/b02eab79b57ba569.png" } ] }, { "benchmark": "frontiercode-1-1", "benchmark_name": "FrontierCode 1.1", "setting": "main", "setting_name": "Main set", "unit": "%", "url": "https://frontierlog.ai/benchmark/frontiercode-1-1", "results": [ { "id": "claude-opus-5-5:frontiercode-1-1:main:2", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "frontiercode-1-1", "benchmark_name": "FrontierCode 1.1", "setting": "main", "setting_name": "Main set", "score": 54.6, "unit": "%", "printed": "54.6%", "reported_by": "anthropic", "run_by": "Cognition", "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5-system-card", "source_title": "Claude Opus 5.5 System Card", "source_date": "2026-09-22", "where": "Section 8.4 'FrontierCode', paragraph ranking Opus 5.5, sentence on Opus 5.5's highest FrontierCode scores, the Main figure", "own": true, "primary": true, "parallel": false, "notes": "medium effort; Claude Code harness", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-20", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [] }, { "id": "gpt-6-astra:frontiercode-1-1:main: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": "frontiercode-1-1", "benchmark_name": "FrontierCode 1.1", "setting": "main", "setting_name": "Main set", "score": 53.3, "unit": "%", "printed": "53.3%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "Coding table, row 'FrontierCode 1.1 Main (score)', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-16", "check": "human", "corroborated": true, "corroborated_on": "2026-10-19", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "53.3", "where": "Table 8.1.A (p.174), row FrontierCode v1.1 (Main), column GPT-6 Astra" }, { "url": "https://www.anthropic.com/claude-opus-5-5", "name": "Claude Opus 5.5", "title": "Introducing Claude Opus 5.5", "lab": "anthropic", "date": "2026-09-22", "printed": "53.3%", "where": "HTML benchmark table under 'Performance and cost-effectiveness', row FrontierCode v1.1 (Main), column GPT-6 Astra" } ], "source_capture": "https://frontierlog.ai/captures/4de8f321fbadf8b8.png" } ] }, { "benchmark": "frontiercode-1-1", "benchmark_name": "FrontierCode 1.1", "setting": "extended", "setting_name": "Extended set", "unit": "%", "url": "https://frontierlog.ai/benchmark/frontiercode-1-1/extended", "results": [ { "id": "claude-opus-5-5:frontiercode-1-1:extended:2", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "frontiercode-1-1", "benchmark_name": "FrontierCode 1.1", "setting": "extended", "setting_name": "Extended set", "score": 65.3, "unit": "%", "printed": "65.3%", "reported_by": "anthropic", "run_by": "Cognition", "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5-system-card", "source_title": "Claude Opus 5.5 System Card", "source_date": "2026-09-22", "where": "Section 8.4 'FrontierCode', paragraph ranking Opus 5.5, sentence on FrontierCode (Extended)", "own": true, "primary": true, "parallel": false, "notes": "medium effort; Claude Code harness", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-20", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [] }, { "id": "gpt-6-astra:frontiercode-1-1:extended: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": "frontiercode-1-1", "benchmark_name": "FrontierCode 1.1", "setting": "extended", "setting_name": "Extended set", "score": 64.5, "unit": "%", "printed": "64.5%", "reported_by": "openai", "run_by": null, "label": "GPT-6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "'Coding' table, row 'FrontierCode 1.1 Extended (score)', column 'GPT-6 Astra' (cell carries footnote 8)", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-20", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/4de8f321fbadf8b8.png" } ] }, { "benchmark": "deepswe-1-1", "benchmark_name": "DeepSWE v1.1", "setting": "default", "setting_name": "", "unit": "%", "url": "https://frontierlog.ai/benchmark/deepswe-1-1", "results": [ { "id": "claude-opus-5-5:deepswe-1-1:default:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "deepswe-1-1", "benchmark_name": "DeepSWE v1.1", "setting": "default", "setting_name": "", "score": 74.2, "unit": "%", "printed": "74.2%", "reported_by": "anthropic", "run_by": null, "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5-system-card", "source_title": "Claude Opus 5.5 System Card", "source_date": "2026-09-22", "where": "Section 8.3 'DeepSWE v1.1', paragraph", "own": true, "primary": true, "parallel": false, "notes": "average of 5 trials", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-17", "check": "human", "corroborated": true, "corroborated_on": "2026-10-19", "corroborated_by": [ { "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/", "name": "Gemini 4 Argon", "title": "Gemini 4 Argon: our next era of frontier intelligence", "lab": "google", "date": "2026-09-30", "printed": "74.2%", "where": "Benchmark table image; row 'DeepSWE v1.1'; column claude-opus-5-5" } ] }, { "id": "gpt-6-astra:deepswe-1-1: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": "deepswe-1-1", "benchmark_name": "DeepSWE v1.1", "setting": "default", "setting_name": "", "score": 74.1, "unit": "%", "printed": "74.1%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "Coding table, row 'DeepSWE v1.1', column 'GPT‑6 Astra'", "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://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/", "name": "Gemini 4 Argon", "title": "Gemini 4 Argon: our next era of frontier intelligence", "lab": "google", "date": "2026-09-30", "printed": "74.1%", "where": "Benchmark table image; row 'DeepSWE v1.1'; column gpt-6-astra" } ], "source_capture": "https://frontierlog.ai/captures/4de8f321fbadf8b8.png" } ] } ], "one": [ { "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "no-tools", "setting_name": "No tools, full set", "unit": "%", "url": "https://frontierlog.ai/benchmark/hle", "result": { "id": "claude-opus-5-5:hle:no-tools:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "no-tools", "setting_name": "No tools, full set", "score": 64.4, "unit": "%", "printed": "64.4", "reported_by": "anthropic", "run_by": null, "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5-system-card", "source_title": "Claude Opus 5.5 System Card", "source_date": "2026-09-22", "where": "Section 8.1 evaluation summary table, row 'Humanity's Last Exam, No tools', column 'Claude Opus 5.5'", "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-haiku-5-5-system-card", "name": "Claude Haiku 5.5 card", "title": "Claude Haiku 5.5 System Card", "lab": "anthropic", "date": "2026-10-07", "printed": "64.4%", "where": "Figure 8.8.1.B HLE (no tools) test-time compute scaling (p.120), Claude Opus 5.5 \"Max\" point label" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "64.4", "where": "Table 8.1.A (p.109), row Humanity's Last Exam / No tools, column Claude Opus 5.5" } ], "source_capture": "https://frontierlog.ai/captures/8a49bd8c526c6c66.png" }, "missing": { "model": "gpt-6-astra", "reported_instead": [ { "kind": "setting", "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "with-tools-unstated", "setting_name": "With tools, set not stated", "unit": "%", "url": "https://frontierlog.ai/benchmark/hle/with-tools-unstated" } ] } }, { "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "with-tools", "setting_name": "With tools, full set", "unit": "%", "url": "https://frontierlog.ai/benchmark/hle/with-tools", "result": { "id": "claude-opus-5-5:hle:with-tools:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "with-tools", "setting_name": "With tools, full set", "score": 67.7, "unit": "%", "printed": "67.7%", "reported_by": "anthropic", "run_by": null, "label": "Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5", "source_title": "Introducing Claude Opus 5.5", "source_date": "2026-09-22", "where": "benchmark table, row 'Humanity's Last Exam', 'with tools', column 'Opus 5.5'", "own": true, "primary": true, "parallel": false, "notes": "max effort", "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-haiku-5-5-system-card", "name": "Claude Haiku 5.5 card", "title": "Claude Haiku 5.5 System Card", "lab": "anthropic", "date": "2026-10-07", "printed": "67.7%", "where": "Figure 8.8.1.A HLE (with tools) reasoning effort scaling (p.119), Claude Opus 5.5 \"Max\" point label" }, { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "67.7", "where": "Table 8.1.A (p.174), row Humanity's Last Exam / With tools, column Claude Opus 5.5" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "67.7", "where": "Table 8.1.A (p.109), row Humanity's Last Exam / With tools, column Claude Opus 5.5" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5", "name": "Claude Sonnet 5.5", "title": "Introducing Claude Sonnet 5.5", "lab": "anthropic", "date": "2026-09-28", "printed": "67.7%", "where": "HTML benchmark table under 'Performance', row Humanity's Last Exam (with tools), column Opus 5.5" } ], "source_capture": "https://frontierlog.ai/captures/ca3ad55d7ec61f3c.png" }, "missing": { "model": "gpt-6-astra", "reported_instead": [ { "kind": "setting", "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "with-tools-unstated", "setting_name": "With tools, set not stated", "unit": "%", "url": "https://frontierlog.ai/benchmark/hle/with-tools-unstated" } ] } }, { "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "with-tools-unstated", "setting_name": "With tools, set not stated", "unit": "%", "url": "https://frontierlog.ai/benchmark/hle/with-tools-unstated", "result": { "id": "gpt-6-astra:hle:with-tools-unstated:1", "model": "gpt-6-astra", "model_name": "GPT-6 Astra", "model_date": "2026-09-03", "event": "https://frontierlog.ai/event/2026-09-03-gpt-6-astra", "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "with-tools-unstated", "setting_name": "With tools, set not stated", "score": 57.2, "unit": "%", "printed": "57.2%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "Academic table, row 'Humanity's Last Exam (w/ tools)', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-16", "check": "human", "corroborated": true, "corroborated_on": "2026-10-19", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "57.2", "where": "Table 8.1.A (p.174), row Humanity's Last Exam / With tools, column GPT-6 Astra" }, { "url": "https://www.anthropic.com/claude-opus-5-5", "name": "Claude Opus 5.5", "title": "Introducing Claude Opus 5.5", "lab": "anthropic", "date": "2026-09-22", "printed": "57.2%", "where": "HTML benchmark table under 'Performance and cost-effectiveness', row Humanity's Last Exam (with tools), column GPT-6 Astra" } ], "source_capture": "https://frontierlog.ai/captures/478a8e3e7a393b2d.png" }, "missing": { "model": "claude-opus-5-5", "reported_instead": [ { "kind": "setting", "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "no-tools", "setting_name": "No tools, full set", "unit": "%", "url": "https://frontierlog.ai/benchmark/hle" }, { "kind": "setting", "benchmark": "hle", "benchmark_name": "Humanity's Last Exam", "setting": "with-tools", "setting_name": "With tools, full set", "unit": "%", "url": "https://frontierlog.ai/benchmark/hle/with-tools" } ] } }, { "benchmark": "swe-bench-pro", "benchmark_name": "SWE-bench Pro", "setting": "unstated", "setting_name": "Set not stated", "unit": "%", "url": "https://frontierlog.ai/benchmark/swe-bench-pro/unstated", "result": { "id": "claude-opus-5-5:swe-bench-pro:unstated:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "swe-bench-pro", "benchmark_name": "SWE-bench Pro", "setting": "unstated", "setting_name": "Set not stated", "score": 89.9, "unit": "%", "printed": "89.9", "reported_by": "anthropic", "run_by": null, "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5-system-card", "source_title": "Claude Opus 5.5 System Card", "source_date": "2026-09-22", "where": "Section 8.1 Evaluation summary, Table 8.1.A, row 'SWE-bench Pro', column 'Claude Opus 5.5'", "own": true, "primary": true, "parallel": false, "notes": "adaptive thinking, max effort; average of 5 trials", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-17", "check": "human", "corroborated": true, "corroborated_on": "2026-10-19", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "89.9", "where": "Table 8.1.A (p.109), row SWE-Bench Pro, column Claude Opus 5.5" } ], "source_capture": "https://frontierlog.ai/captures/8a49bd8c526c6c66.png" }, "missing": { "model": "gpt-6-astra", "reported_instead": [] } }, { "benchmark": "gpqa-diamond", "benchmark_name": "GPQA Diamond", "setting": "default", "setting_name": "", "unit": "%", "url": "https://frontierlog.ai/benchmark/gpqa-diamond", "result": { "id": "gpt-6-astra:gpqa-diamond: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": "gpqa-diamond", "benchmark_name": "GPQA Diamond", "setting": "default", "setting_name": "", "score": 96.0, "unit": "%", "printed": "96.0%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "'Academic' table, column 'GPT‑6 Astra', row 'GPQA Diamond'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-17", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/478a8e3e7a393b2d.png" }, "missing": { "model": "claude-opus-5-5", "reported_instead": [] } }, { "benchmark": "browsecomp", "benchmark_name": "BrowseComp", "setting": "default", "setting_name": "", "unit": "%", "url": "https://frontierlog.ai/benchmark/browsecomp", "result": { "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", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "'Professional' eval table, row 'BrowseComp', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-17", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/8618c81cd990eeed.png" }, "missing": { "model": "claude-opus-5-5", "reported_instead": [] } }, { "benchmark": "arc-agi-2", "benchmark_name": "ARC-AGI-2", "setting": "default", "setting_name": "", "unit": "%", "url": "https://frontierlog.ai/benchmark/arc-agi-2", "result": { "id": "gpt-6-astra:arc-agi-2: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": "arc-agi-2", "benchmark_name": "ARC-AGI-2", "setting": "default", "setting_name": "", "score": 95.0, "unit": "%", "printed": "95.0%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "'Abstract reasoning' table, column 'GPT‑6 Astra', row 'ARC-AGI-2'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-17", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/6917e15279360166.png" }, "missing": { "model": "claude-opus-5-5", "reported_instead": [] } }, { "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "partial", "setting_name": "Full set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1", "result": { "id": "claude-opus-5-5:osworld-2-1:partial:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "partial", "setting_name": "Full set, partial score", "score": 81.8, "unit": "%", "printed": "81.8%", "reported_by": "anthropic", "run_by": null, "label": "Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5", "source_title": "Introducing Claude Opus 5.5", "source_date": "2026-09-22", "where": "benchmark table (HTML), row 'Computer use / OSWorld 2.1', column 'Opus 5.5', the value tagged 'partial'", "own": true, "primary": true, "parallel": false, "notes": "max effort; mean of 5 runs; September 10, 2026 task files; the system card labels it OSWorld 2.0", "caveat": "", "human_verified": false, "human_verified_on": null, "check": "corroborated", "corroborated": true, "corroborated_on": "2026-10-20", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "81.8%", "where": "section 8.13.3 OSWorld 2.0 (p. 206), sentence: 'Claude Opus 5.5 achieved a partial score of 81.8% and a strict pass rate of 48.7%'; also 8.1 Evaluation summary, row 'OSWorld 2.0 (partial/strict)', column Claude Opus 5.5: '81.8/48.7'" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5", "name": "Claude Sonnet 5.5", "title": "Introducing Claude Sonnet 5.5", "lab": "anthropic", "date": "2026-09-28", "printed": "81.8%", "where": "Benchmark table, row 'Computer use / OSWorld 2.1', column Opus 5.5: '81.8%' with 'partial' under it" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "81.8%", "where": "section 8.13.3 OSWorld 2.1 (p. 130), sentence: 'Claude Opus 5.5 scored 81.8% partial and 48.7% strict pass'; also 8.1 Evaluation summary, row 'OSWorld 2.1', column Claude Opus 5.5: '81.8'" } ], "source_capture": "https://frontierlog.ai/captures/ca3ad55d7ec61f3c.png" }, "missing": { "model": "gpt-6-astra", "reported_instead": [ { "kind": "setting", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/offline-partial" }, { "kind": "version", "benchmark": "osworld-2-0-2026-08-08", "benchmark_name": "OSWorld 2.0 (v2026.08.08)", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-0-2026-08-08" } ] } }, { "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "strict", "setting_name": "Full set, strict pass rate", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/strict", "result": { "id": "claude-opus-5-5:osworld-2-1:strict:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "strict", "setting_name": "Full set, strict pass rate", "score": 48.7, "unit": "%", "printed": "48.7%", "reported_by": "anthropic", "run_by": null, "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "source_title": "Claude Sonnet 5.5 System Card", "source_date": "2026-09-28", "where": "section 8.13.3 OSWorld 2.1, text (PDF page 130), the Claude Opus 5.5 strict pass rate", "own": false, "primary": true, "parallel": false, "notes": "max effort; mean of 5 runs; September 10, 2026 task files", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-20", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/ed9b8a21a7f60448.png" }, "missing": { "model": "gpt-6-astra", "reported_instead": [ { "kind": "setting", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/offline-partial" }, { "kind": "version", "benchmark": "osworld-2-0-2026-08-08", "benchmark_name": "OSWorld 2.0 (v2026.08.08)", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-0-2026-08-08" } ] } }, { "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-strict", "setting_name": "Offline set, strict pass rate", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/offline-strict", "result": { "id": "claude-opus-5-5:osworld-2-1:offline-strict:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-strict", "setting_name": "Offline set, strict pass rate", "score": 53.2, "unit": "%", "printed": "53.2%", "reported_by": "anthropic", "run_by": null, "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-haiku-5-5-system-card", "source_title": "Claude Haiku 5.5 System Card", "source_date": "2026-10-07", "where": "section 8.9.3 OSWorld 2.1 offline subset, text (PDF page 128), the Claude Opus 5.5 strict pass rate", "own": false, "primary": true, "parallel": false, "notes": "82-task offline subset, no internet access; max effort; mean of 5 attempts", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-20", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/78b2fca2e8e996c4.png" }, "missing": { "model": "gpt-6-astra", "reported_instead": [ { "kind": "setting", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/offline-partial" }, { "kind": "version", "benchmark": "osworld-2-0-2026-08-08", "benchmark_name": "OSWorld 2.0 (v2026.08.08)", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-0-2026-08-08" } ] } }, { "benchmark": "osworld-2-0-2026-08-08", "benchmark_name": "OSWorld 2.0 (v2026.08.08)", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-0-2026-08-08", "result": { "id": "gpt-6-astra:osworld-2-0-2026-08-08:offline-partial: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": "osworld-2-0-2026-08-08", "benchmark_name": "OSWorld 2.0 (v2026.08.08)", "setting": "offline-partial", "setting_name": "Offline set, partial score", "score": 72.6, "unit": "%", "printed": "72.6%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "Computer Use table, row 'OSWorld 2.0 (v2026.08.08, offline set, partial score)', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-16", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/fbd1c3587f0ca67d.png" }, "missing": { "model": "claude-opus-5-5", "reported_instead": [ { "kind": "version", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "partial", "setting_name": "Full set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1" }, { "kind": "version", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "strict", "setting_name": "Full set, strict pass rate", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/strict" }, { "kind": "version", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-partial", "setting_name": "Offline set, partial score", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/offline-partial" }, { "kind": "version", "benchmark": "osworld-2-1", "benchmark_name": "OSWorld 2.1", "setting": "offline-strict", "setting_name": "Offline set, strict pass rate", "unit": "%", "url": "https://frontierlog.ai/benchmark/osworld-2-1/offline-strict" } ] } }, { "benchmark": "arc-agi-1", "benchmark_name": "ARC-AGI-1", "setting": "unstated", "setting_name": "Set not stated", "unit": "%", "url": "https://frontierlog.ai/benchmark/arc-agi-1/unstated", "result": { "id": "gpt-6-astra:arc-agi-1:unstated:1", "model": "gpt-6-astra", "model_name": "GPT-6 Astra", "model_date": "2026-09-03", "event": "https://frontierlog.ai/event/2026-09-03-gpt-6-astra", "benchmark": "arc-agi-1", "benchmark_name": "ARC-AGI-1", "setting": "unstated", "setting_name": "Set not stated", "score": 98.5, "unit": "%", "printed": "98.5%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "'Abstract reasoning' eval table (HTML), row 'ARC-AGI-1', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-18", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/6917e15279360166.png" }, "missing": { "model": "claude-opus-5-5", "reported_instead": [] } }, { "benchmark": "arc-agi-3", "benchmark_name": "ARC-AGI-3", "setting": "unstated", "setting_name": "Set not stated", "unit": "%", "url": "https://frontierlog.ai/benchmark/arc-agi-3/unstated", "result": { "id": "gpt-6-astra:arc-agi-3:unstated:1", "model": "gpt-6-astra", "model_name": "GPT-6 Astra", "model_date": "2026-09-03", "event": "https://frontierlog.ai/event/2026-09-03-gpt-6-astra", "benchmark": "arc-agi-3", "benchmark_name": "ARC-AGI-3", "setting": "unstated", "setting_name": "Set not stated", "score": 99.9, "unit": "%", "printed": "99.9%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "'Abstract reasoning' eval table (HTML), row 'ARC-AGI-3', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-18", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/6917e15279360166.png" }, "missing": { "model": "claude-opus-5-5", "reported_instead": [] } }, { "benchmark": "chartography", "benchmark_name": "Chartography", "setting": "no-tools", "setting_name": "No tools", "unit": "%", "url": "https://frontierlog.ai/benchmark/chartography", "result": { "id": "claude-opus-5-5:chartography:no-tools:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "chartography", "benchmark_name": "Chartography", "setting": "no-tools", "setting_name": "No tools", "score": 64.4, "unit": "%", "printed": "64.4%", "reported_by": "anthropic", "run_by": null, "label": "Claude Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5-system-card", "source_title": "Claude Opus 5.5 System Card", "source_date": "2026-09-22", "where": "section 8.13.1 Chartography, text (p.200), Claude Opus 5.5 without tools", "own": true, "primary": true, "parallel": false, "notes": "adaptive thinking, max effort; average of 5 runs; Gemini 3.5 Flash grader", "caveat": "", "human_verified": false, "human_verified_on": null, "check": "corroborated", "corroborated": true, "corroborated_on": "2026-10-30", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-sonnet-5-5", "name": "Claude Sonnet 5.5", "title": "Introducing Claude Sonnet 5.5", "lab": "anthropic", "date": "2026-09-28", "printed": "64.4%", "where": "benchmark table, row Visual chart recognition / Chartography, column Opus 5.5, 'no tools'" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "64.4%", "where": "§8.13.1 Chartography prose: 'Claude Opus 5.5 scored 64.4% and 89.0%' (without / with tools)" }, { "url": "https://www.anthropic.com/claude-haiku-5-5-system-card", "name": "Claude Haiku 5.5 card", "title": "Claude Haiku 5.5 System Card", "lab": "anthropic", "date": "2026-10-07", "printed": "64.4", "where": "Figure 8.9.1.A bar label, Claude Opus 5.5, Without tools (value label in a raster figure inside the PDF; not in its extracted text)" } ], "source_capture": "https://frontierlog.ai/captures/f4ec7b7b09abb2fa.png" }, "missing": { "model": "gpt-6-astra", "reported_instead": [] } }, { "benchmark": "chartography", "benchmark_name": "Chartography", "setting": "with-tools", "setting_name": "With tools", "unit": "%", "url": "https://frontierlog.ai/benchmark/chartography/with-tools", "result": { "id": "claude-opus-5-5:chartography:with-tools:1", "model": "claude-opus-5-5", "model_name": "Claude Opus 5.5", "model_date": "2026-09-22", "event": "https://frontierlog.ai/event/2026-09-22-claude-opus-5-5", "benchmark": "chartography", "benchmark_name": "Chartography", "setting": "with-tools", "setting_name": "With tools", "score": 89.0, "unit": "%", "printed": "89.0%", "reported_by": "anthropic", "run_by": null, "label": "Opus 5.5", "source": "https://www.anthropic.com/claude-opus-5-5", "source_title": "Introducing Claude Opus 5.5", "source_date": "2026-09-22", "where": "benchmark table, row 'Visual chart recognition, Chartography', column 'Opus 5.5', 'with tools'", "own": true, "primary": true, "parallel": false, "notes": "adaptive thinking, max effort", "caveat": "", "human_verified": false, "human_verified_on": null, "check": "corroborated", "corroborated": true, "corroborated_on": "2026-10-30", "corroborated_by": [ { "url": "https://www.anthropic.com/claude-opus-5-5-system-card", "name": "Claude Opus 5.5 card", "title": "Claude Opus 5.5 System Card", "lab": "anthropic", "date": "2026-09-22", "printed": "89.0%", "where": "§8.13.1 Chartography prose: 'Claude Opus 5.5 achieved a score of 64.4% without tools and 89.0% with tools'" }, { "url": "https://www.anthropic.com/claude-sonnet-5-5-system-card", "name": "Claude Sonnet 5.5 card", "title": "Claude Sonnet 5.5 System Card", "lab": "anthropic", "date": "2026-09-28", "printed": "89.0%", "where": "§8.13.1 Chartography prose: 'Claude Opus 5.5 scored 64.4% and 89.0%'" }, { "url": "https://www.kimi.ai/blog/kimi-k3-1-pro", "name": "Kimi K3.1 Pro", "title": "Kimi K3.1 Pro Tech Blog: Advancing Open Intelligence", "lab": "moonshot", "date": "2026-10-24", "printed": "89.0", "where": "benchmark table (rendered), row 'ChartoGraphy w/ python', column 'Claude Opus 5.5 (max, with fallback)'" } ] }, "missing": { "model": "gpt-6-astra", "reported_instead": [] } }, { "benchmark": "sec-bench-pro-no-release", "benchmark_name": "SEC-bench Pro, no release named", "setting": "default", "setting_name": "", "unit": "%", "url": "https://frontierlog.ai/benchmark/sec-bench-pro-no-release", "result": { "id": "gpt-6-astra:sec-bench-pro-no-release: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": "sec-bench-pro-no-release", "benchmark_name": "SEC-bench Pro, no release named", "setting": "default", "setting_name": "", "score": 85.4, "unit": "%", "printed": "85.4%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "Cybersecurity eval table (HTML; columns GPT‑6 Astra, GPT‑5.6 Sol, Claude Fable 5.1, ...), row 'SEC-Bench Pro', column 'GPT‑6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "", "caveat": "", "human_verified": false, "human_verified_on": null, "check": "corroborated", "corroborated": true, "corroborated_on": "2026-10-30", "corroborated_by": [ { "url": "https://storage.googleapis.com/deepmind-media/gemini/gemini_4_carbon_model_evaluation.pdf", "name": "Gemini 4 Carbon card", "title": "Gemini 4 Carbon Model evaluation: approach, methodology & results", "lab": "google", "date": "2026-10-28", "printed": "85.4%", "where": "page 5 'Results' table (raster image), row 'SEC-bench Pro', column 'GPT-6 Astra' (Google restating; footnote: competitor numbers taken from their published system cards and technical reports)" }, { "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/", "name": "Gemini 4 Carbon", "title": "Gemini 4 Carbon: our next frontier model", "lab": "google", "date": "2026-10-28", "printed": "85.4%", "where": "SEC-bench Pro bar chart (Pass@1), value label on bar 'GPT-6 Astra'" } ] }, "missing": { "model": "claude-opus-5-5", "reported_instead": [] } }, { "benchmark": "agents-last-exam", "benchmark_name": "Agents' Last Exam v1", "setting": "unstated", "setting_name": "Measure not stated", "unit": "%", "url": "https://frontierlog.ai/benchmark/agents-last-exam/unstated", "result": { "id": "gpt-6-astra:agents-last-exam:unstated:1", "model": "gpt-6-astra", "model_name": "GPT-6 Astra", "model_date": "2026-09-03", "event": "https://frontierlog.ai/event/2026-09-03-gpt-6-astra", "benchmark": "agents-last-exam", "benchmark_name": "Agents' Last Exam v1", "setting": "unstated", "setting_name": "Measure not stated", "score": 59.3, "unit": "%", "printed": "59.3%", "reported_by": "openai", "run_by": null, "label": "GPT‑6 Astra", "source": "https://openai.com/index/gpt-6-astra", "source_title": "GPT-6 Astra: A new generation of intelligence", "source_date": "2026-09-03", "where": "'Computer Use' eval table, row 'Agents' Last Exam', column 'GPT-6 Astra'", "own": true, "primary": true, "parallel": false, "notes": "best of any effort; harness not stated", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-30", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [] }, "missing": { "model": "claude-opus-5-5", "reported_instead": [] } }, { "benchmark": "mmmu-pro", "benchmark_name": "MMMU-Pro", "setting": "vision", "setting_name": "Vision setting", "unit": "%", "url": "https://frontierlog.ai/benchmark/mmmu-pro", "result": { "id": "gpt-6-astra:mmmu-pro:vision:5", "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": "mmmu-pro", "benchmark_name": "MMMU-Pro", "setting": "vision", "setting_name": "Vision setting", "score": 86.7, "unit": "%", "printed": "86.7%", "reported_by": "openai", "run_by": null, "label": "GPT-6 Astra (Max)", "source": "https://openai.com/index/introducing-gpt-6-1-terra", "source_title": "Introducing GPT-6.1 Terra", "source_date": "2026-10-26", "where": "\"MMMU-Pro (Vision)\" chart (score vs cost per task; no value labels): point 'GPT-6 Astra', effort 'Max', read from the point's aria-label (load the page in a browser and hover or inspect the SVG point)", "own": false, "primary": true, "parallel": false, "notes": "max effort", "caveat": "", "human_verified": true, "human_verified_on": "2026-11-08", "check": "human", "corroborated": false, "corroborated_on": null, "corroborated_by": [], "source_capture": "https://frontierlog.ai/captures/0b120dcee7cbdefb.png" }, "missing": { "model": "claude-opus-5-5", "reported_instead": [] } } ] } }