---
title: "$τ$-bench: A Benchmark for Tool-Agent-User Interaction"
url: https://stacklist.com/card/373a54aa-196d-40c3-a3ca-4fb60cb05b43
source_url: "https://arxiv.org/abs/2406.12045"
stack: https://stacklist.com/c/technology/stack/29e01ac8-6202-4abf-ab14-a0613026186b
summary: "τ-bench is a benchmark for evaluating language agent interactions with simulated human users in real-world domains, testing agents' ability to use domain-specific API tools and follow policy guidelines. The paper introduces a pass^k reliability metric and finds that even state-of-the-art models like GPT-4o succeed on less than 50% of tasks, highlighting the need for more consistent and rule-following agent methods."
tags: "benchmark, language-agents, tool-use, human-agent-interaction, evaluation, function-calling, reliability"
key_entities: "τ-bench (concept), Shunyu Yao (person), Noah Shinn (person), Pedram Razavi (person), Karthik Narasimhan (person), GPT-4o (technology), arXiv (organization), pass^k metric (concept), function-calling agents (concept)"
classification: "reference"
content_hash: "sha256:7ef81f098b387dbd454e75a834ee03961b3c1c3842b8da439bba72c34c190778"
acp_version: "0.2"
token_counts_approximate: 988
visibility: public
agent_accessible: true
status: "final"
---

# $τ$-bench: A Benchmark for Tool-Agent-User Interaction

--> Computer Science > Artificial Intelligence arXiv:2406.12045 (cs) [Submitted on 17 Jun 2024] Title: $τ$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains Authors: Shunyu Yao , Noah Shinn , Pedram Razavi , Karthik Narasimhan View a PDF of the paper titled $\tau$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains, by Shunyu Yao and 3 other authors View PDF Abstract: Existing benchmarks do not test language agents on their interaction with human users or ability to follow domain-specific rules, both of which are vital for deploying them in real world applications. We propose $\tau$-bench, a benchmark emulating dynamic conversations between a user (simulated by language models) and a language agent provided with domain-specific API tools and policy guidelines. We employ an efficient and faithful evaluation process that compares the database state at the end of a conversation with the annotated goal state. We also propose a new metric (pass^k) to evaluate the reliability of agent behavior over multiple trials. Our experiments show that even state-of-the-art function calling agents (like gpt-4o) succeed on <50% of the tasks, and are quite inconsistent (pass^8 <25% in retail). Our findings point to the need for methods that can improve the ability of agents to act consistently and follow rules reliably. Subjects: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL) Cite as: arXiv:2406.12045 [cs.AI]   (or arXiv:2406.12045v1 [cs.AI] for this version)   https://doi.org/10.48550/arXiv.2406.12045 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Karthik Narasimhan [ view email ] [v1] Mon, 17 Jun 2024 19:33:08 UTC (647 KB) Full-text links: Access Paper: View a PDF of the paper titled $\tau$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains, by Shunyu Yao and 3 other authors View PDF TeX Source view license Current browse context: cs.AI < prev   |   next > new | recent | 2024-06 Change to browse by: cs cs.CL References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) scite.ai Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle Gotit.pub ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle TXYZ.AI ( What is TXYZ.AI? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs . Which authors of this paper are endorsers? | Disable MathJax ( What is MathJax? )