Benchmark
Tau-bench
What is Tau-bench?
τ-bench: A benchmark for tool-agent-user interaction in real-world domains. Tests language agents' ability to interact with users and follow domain-specific…
- Released
- 2024-06-17
- Modality
- text
- Evaluates
- reasoning, general, agents, tool calling
- Openness
- unknown
- Importer
- LLM Stats
- Reported scores
- 6
Source provenance
- Original evidence https://api.zeroeval.com/leaderboard/benchmarks/tau-bench?top_n=500
Tau-bench results and reported scores
LLM Stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| GLM-4.7 | Z.ai | 0.874 | 2025-12-22 | evidence |
| GLM-4.7-Flash | Z.ai | 0.795 | 2026-01-19 | evidence |
| MiMo-V2-Flash | Xiaomi | 0.803 | 2025-12-16 | evidence |
| MiniMax M2 | MiniMax | 0.772 | 2025-10-27 | evidence |
| o3 | OpenAI | 0.63 | 2025-04-16 | evidence |
| Step-3.5-Flash | StepFun | 0.882 | 2026-02-02 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.