$τ^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment

Victor Barres, Honghua Dong, Soham Ray, Xujie Si, Karthik R Narasimhan
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:6736-6782, 2026.

Abstract

Existing benchmarks for conversational AI agents simulate single-control environments, where only the AI agent can use tools to interact with the world, while the user remains a passive information provider. This differs from real-world scenarios like technical support, where users need to actively participate in modifying the state of the (shared) world. In order to address this gap, we introduce $\tau^2$-bench, with four key contributions: (1) A novel Telecom dual-control domain modeled as a Dec-POMDP, where both agent and user make use of tools to act in a shared, dynamic environment that tests both agent coordination and communication; (2) A compositional task generator that programmatically creates diverse, verifiable tasks from atomic components, ensuring domain coverage and controlled complexity; (3) A reliable user simulator tightly coupled with the environment, whose behavior is constrained by tools and observable states, improving simulation fidelity; (4) Fine-grained analysis of agent performance through multiple ablations including separating errors arising from reasoning vs communication/coordination. In particular, our experiments show significant performance drops when agents shift from no-user to dual-control, highlighting the challenges of guiding users. Overall, $\tau^2$-bench provides a controlled testbed for agents that must both reason effectively and guide user actions. Code, data, and leaderboard are available at https://taubench.com/.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-barres26a, title = {$τ^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment}, author = {Barres, Victor and Dong, Honghua and Ray, Soham and Si, Xujie and Narasimhan, Karthik R}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {6736--6782}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/barres26a/barres26a.pdf}, url = {https://proceedings.mlr.press/v306/barres26a.html}, abstract = {Existing benchmarks for conversational AI agents simulate single-control environments, where only the AI agent can use tools to interact with the world, while the user remains a passive information provider. This differs from real-world scenarios like technical support, where users need to actively participate in modifying the state of the (shared) world. In order to address this gap, we introduce $\tau^2$-bench, with four key contributions: (1) A novel Telecom dual-control domain modeled as a Dec-POMDP, where both agent and user make use of tools to act in a shared, dynamic environment that tests both agent coordination and communication; (2) A compositional task generator that programmatically creates diverse, verifiable tasks from atomic components, ensuring domain coverage and controlled complexity; (3) A reliable user simulator tightly coupled with the environment, whose behavior is constrained by tools and observable states, improving simulation fidelity; (4) Fine-grained analysis of agent performance through multiple ablations including separating errors arising from reasoning vs communication/coordination. In particular, our experiments show significant performance drops when agents shift from no-user to dual-control, highlighting the challenges of guiding users. Overall, $\tau^2$-bench provides a controlled testbed for agents that must both reason effectively and guide user actions. Code, data, and leaderboard are available at https://taubench.com/.} }
Endnote
%0 Conference Paper %T $τ^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment %A Victor Barres %A Honghua Dong %A Soham Ray %A Xujie Si %A Karthik R Narasimhan %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-barres26a %I PMLR %P 6736--6782 %U https://proceedings.mlr.press/v306/barres26a.html %V 306 %X Existing benchmarks for conversational AI agents simulate single-control environments, where only the AI agent can use tools to interact with the world, while the user remains a passive information provider. This differs from real-world scenarios like technical support, where users need to actively participate in modifying the state of the (shared) world. In order to address this gap, we introduce $\tau^2$-bench, with four key contributions: (1) A novel Telecom dual-control domain modeled as a Dec-POMDP, where both agent and user make use of tools to act in a shared, dynamic environment that tests both agent coordination and communication; (2) A compositional task generator that programmatically creates diverse, verifiable tasks from atomic components, ensuring domain coverage and controlled complexity; (3) A reliable user simulator tightly coupled with the environment, whose behavior is constrained by tools and observable states, improving simulation fidelity; (4) Fine-grained analysis of agent performance through multiple ablations including separating errors arising from reasoning vs communication/coordination. In particular, our experiments show significant performance drops when agents shift from no-user to dual-control, highlighting the challenges of guiding users. Overall, $\tau^2$-bench provides a controlled testbed for agents that must both reason effectively and guide user actions. Code, data, and leaderboard are available at https://taubench.com/.
APA
Barres, V., Dong, H., Ray, S., Si, X. & Narasimhan, K.R.. (2026). $τ^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:6736-6782 Available from https://proceedings.mlr.press/v306/barres26a.html.

Related Material