tau2-bench
τ-Bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains
tau2-bench scores 5.7/10 (Solid) in Benchmarks & Resources.
Kind Resource Language Python License MIT Stars 1.5k★ Repo sierra-research/tau2-bench
Quality
Solid 5.7
Adoption 30
Activity 63
Maturity 82
Community 40
Capability 50
Show the math
Overall 5.7/10
value = 5.7/10
score = value
Adoption 3.7/10
value = 29.5/100
score = 1 + 9 * value / 100
Activity 6.7/10
value = 63.3/100
score = 1 + 9 * value / 100
Maturity 8.4/10
value = 81.8/100
score = 1 + 9 * value / 100
Community 4.6/10
value = 39.9/100
score = 1 + 9 * value / 100
Capability 5.5/10
value = 50/100
score = 1 + 9 * value / 100
Key metrics
1.5k★ GitHub stars
61/90d Recent commits
8 devs/90d Recent contributors
1.1y Project age
active Last commit
Gotchas
No gotchas documented yet. Contribute one if you know a constraint we missed.
benchmark tool-use multimodal agents