Benchmarks / Agentic
τ^τ-bench (Hyper-τ), release v1 (Sierra)
A coding agent builds a working customer-service agent from realistic evidence (policies, transcripts, recordings, a client API), and is scored by how well that agent then serves simulated customers on 53 held-out tasks. Maintained by Sierra. Score: Overall: mean reward across all 53 release tasks of the agent the Developer built (τ³-bench inner loop).
τ^τ-bench (Hyper-τ), release v1 (Sierra)
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Opus 5 | max | 23.9% |
| 2 | GPT-5.6 Sol | xhigh | 22.0% |
| 3 | GPT-5.6 Terra | xhigh | 18.0% |
| 4 | Kimi K3 | max | 17.9% |
| 5 | Sonnet 5 | max | 14.9% |