τ²-bench (tau2-bench)

Measures: An agent serves a simulated user over several turns in the airline, retail and telecom domains (in telecom the user also acts on a shared state, dual control), using tools and following policy; scored as Pass^1 task success.

Maintainer: Sierra (sierra-research)
Introduced: 2025
Status: Superseded
Saturation: Saturated
Typical provenance: Mixed
In scoring set: No

Paper Official leaderboard

Leaderboard

Show independent results only

Every result we hold, whoever ran it. Only independent and third-party rows count towards a score.

No results to show.