{
  "id": "tau-bench",
  "code": "PTL-0204",
  "term": "τ-bench",
  "aliases": [
    "tau-bench",
    "TAU-bench",
    "pass^k"
  ],
  "category": "benchmarks",
  "definition": "τ-bench is a benchmark in which an agent with API tools and a written domain policy, for example for airline or retail customer service, converses with a simulated user and is judged by whether the final database state is correct.",
  "description": "It tests tool use, policy following, and conversation together, as in real deployments. It introduced pass^k, the probability that an agent succeeds on all of k repeated trials of the same task, which exposes how inconsistent agents can be even when they often succeed once.",
  "example": null,
  "broader": [
    "benchmark"
  ],
  "narrower": [],
  "related": [
    "function-calling",
    "ai-agent",
    "pass-at-k"
  ],
  "introduced": 2024,
  "added": "2026-10-10",
  "sources": [
    {
      "title": "τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains",
      "authors": "Yao et al.",
      "year": 2024,
      "url": "https://arxiv.org/abs/2406.12045"
    }
  ],
  "url": "https://protologue.com/t/tau-bench/",
  "citation": "Protologue. (2026). τ-bench. In Protologue: A Taxonomy of Prompting and LLM Techniques (v1.4.0, PTL-0204). https://protologue.com/t/tau-bench/"
}