Tau3-Bench Airline — leaderboard

τ³-bench airline domain: multimodal, knowledge-aware agentic customer service tasks. New domain in τ³-bench. From Sierra AI.

Metric: Pass@1 (%). Source: taubench.com. Status: saturation imminent. 9 models tracked.

Top models

#ModelScore
1GLM-5.287.5
2Claude Opus 4.584
3GPT-5.283
4GLM-582.5
5Gemini 3 Flash82.5
6Qwen 3.5 397B A17B81.5
7Gemini 3 Pro80.5
8Claude Sonnet 4.572

Interactive version: theaggregate.ai/benchmark?slug=tau3-bench-airline · How the rankings work · Data refreshed daily, snapshot 2026-07-22.