List your company

Agent Performance on Tau-bench Retail

60.31Avg@4NoisyAgent
34.3141.0647.8154.56May 26, 2026

Evaluation results

MethodLinks
NoisyAgentbackbone=Qwen3-32B2026.0560.3186.84
GSPObackbone=Qwen3-32B2026.0558.5584.21
GRPObackbone=Qwen3-32B2026.0558.1183.33
DAPObackbone=Qwen3-32B2026.0556.5880.7
Qwen3-32Bbackbone=Qwen3-32B2026.0549.1272.81
NoisyAgentbackbone=Qwen3-8B2026.0547.5977.19
GSPObackbone=Qwen3-8B2026.0546.4974.56
GRPObackbone=Qwen3-8B2026.0546.0573.68
DAPObackbone=Qwen3-8B2026.0544.5271.05
Qwen3-8Bbackbone=Qwen3-8B2026.0535.3159.65