Anthropic Claude Haiku 5.5 (max)39.2% anthropic.com for Claude Haiku 5.5 (max) on Terminal-Bench 4.0 (opens in a new tab) Archived copy (opens in a new tab) Terminal-Bench Accuracy Reported October 7, 2026 System card Anthropic Self-reported Record for Claude Haiku 5.5 on Terminal-Bench 4.0 20018 Google (DeepMind) Gemini 4 Argon (highest thinking settings)57.4% deepmind.google for Gemini 4 Argon (highest thinking settings) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 30, 2026 Evaluation methodology Google DeepMind Self-reported Record for Gemini 4 Argon on Terminal-Bench 4.0 30011 Anthropic Claude Sonnet 5.5 (max)70.6% anthropic.com for Claude Sonnet 5.5 (max) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 28, 2026 System card Anthropic Self-reported Record for Claude Sonnet 5.5 on Terminal-Bench 4.0 20016 Anthropic Claude Sonnet 5.5 (high)43.0% anthropic.com for Claude Sonnet 5.5 (high) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 28, 2026 Launch post Anthropic Self-reported Record for Claude Sonnet 5.5 on Terminal-Bench 4.0 20016 Anthropic Claude Sonnet 5.5 (xhigh)61.5% anthropic.com for Claude Sonnet 5.5 (xhigh) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 28, 2026 Launch post Anthropic Self-reported Record for Claude Sonnet 5.5 on Terminal-Bench 4.0 20016 Anthropic Claude Sonnet 5.5 (low)20.0% anthropic.com for Claude Sonnet 5.5 (low) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 28, 2026 Launch post Anthropic Self-reported Record for Claude Sonnet 5.5 on Terminal-Bench 4.0 20016 Anthropic Claude Sonnet 5.5 (medium)28.8% anthropic.com for Claude Sonnet 5.5 (medium) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 28, 2026 Launch post Anthropic Self-reported Record for Claude Sonnet 5.5 on Terminal-Bench 4.0 20016 Anthropic Claude Opus 5.5 (max)64.8% www-cdn.anthropic.com for Claude Opus 5.5 (max) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 22, 2026 System card Anthropic Self-reported Record for Claude Opus 5.5 on Terminal-Bench 4.0 20015 Anthropic Claude Opus 5.5 (xhigh)66.4% anthropic.com for Claude Opus 5.5 (xhigh) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 22, 2026 Launch post Anthropic Self-reported Record for Claude Opus 5.5 on Terminal-Bench 4.0 20015 SpaceXAI Grok 4.7 (xhigh)37.6% x.ai for Grok 4.7 (xhigh) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 21, 2026 Launch post SpaceXAI Self-reported Record for Grok 4.7 on Terminal-Bench 4.0 40003 OpenAI GPT-6 Astra 57.9% openai.com for GPT-6 Astra on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 3, 2026 Launch post OpenAI Self-reported Record for GPT-6 Astra on Terminal-Bench 4.0 10005 Google (DeepMind) Gemini 3.8 Flash 19.1% deepmind.google for Gemini 3.8 Flash on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 2, 2026 Model card Google DeepMind Self-reported Terminal-Bench Team Record for Gemini 3.8 Flash on Terminal-Bench 4.0 30004 Anthropic Claude Fable 5.1 (low)40.2% anthropic.com for Claude Fable 5.1 (low) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 1, 2026 Launch post Anthropic Self-reported Record for Claude Fable 5.1 on Terminal-Bench 4.0 20004 Anthropic Claude Fable 5.1 (xhigh)51.3% anthropic.com for Claude Fable 5.1 (xhigh) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 1, 2026 Launch post Anthropic Self-reported Record for Claude Fable 5.1 on Terminal-Bench 4.0 20004 Anthropic Claude Fable 5.1 (high)49.4% anthropic.com for Claude Fable 5.1 (high) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 1, 2026 Launch post Anthropic Self-reported Record for Claude Fable 5.1 on Terminal-Bench 4.0 20004 Anthropic Claude Fable 5.1 55.8% anthropic.com for Claude Fable 5.1 on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 1, 2026 Launch post Anthropic Self-reported Record for Claude Fable 5.1 on Terminal-Bench 4.0 20004 Anthropic Claude Fable 5.1 (medium)43.4% anthropic.com for Claude Fable 5.1 (medium) on Terminal-Bench 4.0 (opens in a new tab) Terminal-Bench Accuracy Reported September 1, 2026 Launch post Anthropic Self-reported Record for Claude Fable 5.1 on Terminal-Bench 4.0 20004