Tau-Bench Retail

Ï„-Bench Retail - function calling ability benchmark in a retail domain. Measures tool use with developer-defined function schemas.

Category

General Agents

Max Score

100.0

Score Type

percent

Active

Yes

Model Scores

Scores for Tau-Bench Retail

Model Score Date Verified
GPT-OSS-120B
87.84%
08.08.2025 Verified
GPT-OSS-20B
54.80
05.08.2025 Verified
Qwen3-Next-80B-A3B-Thinking
100.00%
10.09.2025 Verified