Benchmarks
Benchmarks

τ²-Bench Retail

Agentsunit % · normalized over [30, 99]

τ²-Bench retail domain — dual-control tool-agent-user tasks in a simulated e-commerce/retail support setting; a harder successor to tau-bench, distinct scale.

#ModelSourceScoreNormalized
Score distribution
10 tracked results across the normalization window
3099
Score vs. parameters
Open-weights models, log-x params
1B10B100B1000B