{"slug":"tau2-bench","name":"tau2-bench","category":"eval","categoryName":"Evaluation, red-teaming & guardrails","audience":["ai-era"],"primary":"ai-era","openClosed":"open","pricing":"open-source","license":"MIT","maintained":"yes","officialUrl":"https://www.taubench.com","repo":"https://github.com/sierra-research/tau2-bench","latestVersion":"v1.0.1","latestRelease":"2026-07-22","stars":2160,"starsAsOf":"2026-10-04","description":"τ-Bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains","descriptionZh":"τ²-bench 双控制环境基准","callMethods":["CLI"],"platforms":[],"addedAt":"2026-10-10","updatedAt":"2026-10-04","topics":["ai","benchmark","conversational-agents","language-model-agent","llm"],"language":"Python","lastPush":"2026-09-28","formerly":null,"packages":[],"recentVersions":[{"version":"v1.0.1","date":"2026-07-22","prerelease":false,"url":"https://github.com/sierra-research/tau2-bench/releases/tag/v1.0.1","source":"GitHub Release"},{"version":"v1.0.0","date":"2026-03-18","prerelease":false,"url":"https://github.com/sierra-research/tau2-bench/releases/tag/v1.0.0","source":"GitHub Release"},{"version":"v0.2.0","date":"2025-10-06","prerelease":false,"url":"https://github.com/sierra-research/tau2-bench/releases/tag/v0.2.0","source":"GitHub Release"},{"version":"v0.1.3","date":"2025-08-26","prerelease":false,"url":"https://github.com/sierra-research/tau2-bench/releases/tag/v0.1.3","source":"GitHub Release"},{"version":"v0.1.2","date":"2025-07-17","prerelease":false,"url":"https://github.com/sierra-research/tau2-bench/releases/tag/v0.1.2","source":"GitHub Release"}],"alternatives":["opencompass","osworld","browsergym","promptfoo","ragas","swe-bench","giskard-oss","gorilla"],"stacks":[],"scenarios":["evaluate-and-monitor"],"problems":[],"reasonZh":"MiniMax 交错思维丢失 Tau² 87→64","sources":["profile","priority","oss"],"record":"https://aimap.page/api/v1/software/tau2-bench.json","pages":{"en":"https://aimap.page/en/software/tau2-bench/","zh-hans":"https://aimap.page/zh-hans/software/tau2-bench/"}}