List your company

Per-step Tool Selection Prediction on τ-bench airline multi-step

100Success RateQwen2.5-3B
46.9660.7374.588.27Jun 15, 2026

Evaluation results

MethodLinks
Qwen2.5-3Bstatus=ceiling2026.06100199.8
Llama-3.1-8Bstatus=0 fails2026.06100——
Qwen2.5-7Bstatus=noise2026.06950.38723.6
Qwen2.5-0.5Bstatus=inform.2026.06550.98565.3
Qwen2.5-1.5Bstatus=inform.2026.06490.99897