← τ-bench leaderboards

Airline · Standard · pass^4

τ-bench · 8 models

RankModelScore
1GPT-5.272
2Claude Opus 4.570
2GLM-570
4Gemini 3 Flash68
4Qwen3.5-397B-A17B68
6Gemini 3 Pro66
7Claude Sonnet 4.548
8GPT-5.222