Revision History

Auto-saved before update

Correction: dated Opus 5.5's first place on Artificial Analysis Intelligence Index v5.0 (October 27, 2026) and added that Gemini 4 Carbon (high) took first place at 59 on October 28

Update: fast mode status and platforms as of October 2026, Haiku 5.5 release, October 6 Cyber Verification Program tiers and CyScenarioBench results, distillation defenses, GitHub Copilot and Kiro availability, later Artificial Analysis/Epoch/Vals leaderboard results, OpenRouter usage data

Improve internal links (2026-10-13)

Verification fixes: OSWorld label history (launch post edited 28 Sep); system-card row keeps card label; FrontierCode sentence logic

Correction: OSWorld 2.0 relabelled 2.1 with attribution; Sonnet 5.5 released Sep 28; Fable 5.1 FrontierCode 52.8% vs 50.9% conflict attributed; comparability claim re-attributed to Artificial Analysis; CVP status updated; links to new pages

Correction: reference 6 date is September 12, 2026; link "We Must Pace the Frontier"

Verification fixes: safeguards wording now 'similar' and attributed; benchmark provenance (Cognition, Cursor, Artificial Analysis, Zapier); FrontierCode effort note is Anthropic's; research-task description; HealthBench length-adjusted note; re-run/version comparability caveat; AA-Briefcase rounding; second sandbox evaluation caveat

No summary provided