Benchmarks

τ³-Banking

Agentsunit % · normalized over [0, 45]

τ³-Bench banking domain (Artificial Analysis) — a newer, substantially harder generation of dual-control tool-agent-user tasks than τ²-Bench; scores are much lower across the field.

#ModelSourceScoreNormalized

Score distribution

4 tracked results across the normalization window

045

Score vs. parameters

Open-weights models, log-x params