OpenAI GPT-5.6 Sol (xhigh)46.9% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for GPT-5.6 Sol (xhigh) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated July 22, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for GPT-5.6 Sol on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 10010 Moonshot AI Kimi K3 (max)37.1% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Kimi K3 (max) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated July 24, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Kimi K3 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 120003 Z.ai GLM-5.2 (xhigh)37.1% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for GLM-5.2 (xhigh) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated July 24, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for GLM-5.2 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 150004 Thinking Machines Lab Inkling (max)25.0% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Inkling (max) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated July 24, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Inkling on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 160001 Anthropic Claude Opus 4.8 (max)39.7% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Claude Opus 4.8 (max) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated July 23, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Claude Opus 4.8 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 20011 Anthropic Claude Fable 5 (max)39.7% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Claude Fable 5 (max) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated July 23, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Claude Fable 5 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 20012 Anthropic Claude Opus 5 (max)48.7% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Claude Opus 5 (max) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated August 3, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Claude Opus 5 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 20014 SpaceXAI Grok 4.5 (high)47.9% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Grok 4.5 (high) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated July 24, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Grok 4.5 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 40001 Meta AI (originally Facebook AI Research) Muse Spark 1.1 (xhigh)40.5% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Muse Spark 1.1 (xhigh) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported August 4, 2026 Evaluated July 23, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Muse Spark 1.1 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 80002 Anthropic Claude Opus 4.7 (max)40.2% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Claude Opus 4.7 (max) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported July 23, 2026 Evaluated July 23, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Claude Opus 4.7 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 20010 OpenAI GPT-5.5 (xhigh)44.6% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for GPT-5.5 (xhigh) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported July 22, 2026 Evaluated July 22, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for GPT-5.5 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 10008 OpenAI GPT-5.4 (xhigh)39.4% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for GPT-5.4 (xhigh) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported May 6, 2026 Evaluated May 6, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for GPT-5.4 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 10007 Anthropic Claude Opus 4.6 (max)27.3% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Claude Opus 4.6 (max) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported May 6, 2026 Evaluated May 6, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Claude Opus 4.6 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 20008 OpenAI GPT-5.2 (high)32.2% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for GPT-5.2 (high) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported May 5, 2026 Evaluated May 5, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for GPT-5.2 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 10012 Anthropic Claude Opus 4.5 (high)24.7% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Claude Opus 4.5 (high) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported May 5, 2026 Evaluated May 5, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Claude Opus 4.5 on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 20007 Google (DeepMind) Gemini 3.1 Pro (high)26.0% sierra-tau-bench-public.s3.us-west-2.amazonaws.com for Gemini 3.1 Pro (high) on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading (opens in a new tab) Archived copy (opens in a new tab) Task success rate Reported May 5, 2026 Evaluated May 5, 2026 Benchmark leaderboard Sierra Independent Sierra Research Record for Gemini 3.1 Pro on τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading 30005