Elamur Research⌕
☰
home / papers / 2026 / TEA-Bench: A Systematic Benchmarking of Tool-enhanced Emotional Support Dialogue Agent

TEA-Bench: A Systematic Benchmarking of Tool-enhanced Emotional Support Dialogue Agent

Paper · 2026 · Xingyu Sui, Yanyan Zhao, Yulin Hu et al.

Summary

Most emotional support conversation (ESC) systems and benchmarks only measure empathetic wording in text-only settings. Practical advice therefore goes unchecked against real-world context, and models may invent details such as the time of day or the user's location. The authors present TEA-Bench, which they describe as the first interactive benchmark for tool-augmented agents in ESC. It adapts scenarios from ExTES, adds hidden spatiotemporal context, and gives agents an MCP-style tool environment with tools for time, location, maps, weather, music and news. A simulated user reacts to the agent's replies, including replies with fabricated content. Process-level metrics score both support quality and factual grounding across turns.

Across nine LLMs, tools generally improved support and reduced hallucination, but the benefit depended on model capacity. Strong models called tools sparingly and to good effect, while weak models gained little. The authors also release TEA-Dialog. Fine-tuning on it helped in familiar scenarios but transferred poorly to unseen ones. The work shows that grounding is a central requirement for trustworthy support agents.

Key points

  • TEA-Scenarios: 81 scenarios built from ExTES through a four-stage pipeline. Only situations described in more than 30 words were kept. An LLM then inferred latent attributes for each scenario: local time, city-level location and place type.
  • Information asymmetry: the agent sees its own tool calls and outputs, while the user and evaluators see only the dialogue text. Claims are counted as factual only if they trace back to the user's statements or to tool observations.
  • Models evaluated: nine LLMs, four closed-source and five open-source.
    • Hallucination reduction grew as tool calls increased.
    • Mid-capability models needed more calls than strong models to reach similar grounding.
  • TEA-Dialog: 365 dialogues, each with a TEA-Score above 80, no detected hallucinations, and a human check.
    • Early turns favour utility and context tools.
    • Middle turns favour maps and weather.
    • Later turns favour personalised tools such as music and news.
  • Fine-tuning experiment: Qwen3-8B and Qwen3-14B were fine-tuned on dialogues from the first 60 scenarios, producing Qwen3-8B-TEA and Qwen3-14B-TEA.
    • Information and Effectiveness scores rose.
    • Hallucination fell on in-domain scenarios but rose sharply out-of-domain.
    • The 14B model showed a smaller in-domain vs. out-of-domain gap than the 8B model.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Harbin Institute of TechnologyHarbin · China158 works · 28,841 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksPsychology · 1,552 worksExperimental and Cognitive Psychology · 870 works

🧭 Topics 2Research subjects and languages the work is about.Sentiment & emotion · 972 worksDialogue systems · 624 works

Cite

Cite as Xingyu Sui, Yanyan Zhao, Yulin Hu, Jiahe Guo, Weixiang Zhao and Bing Qin. 2026. TEA-Bench: A Systematic Benchmarking of Tool-enhanced Emotional Support Dialogue Agent. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.2152

@inproceedings{sui2026tea,
  title = {{TEA-Bench: A Systematic Benchmarking of Tool-enhanced Emotional Support Dialogue Agent}},
  author = {Xingyu Sui, Yanyan Zhao, Yulin Hu, Jiahe Guo, Weixiang Zhao and Bing Qin},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.2152},
  url = {https://doi.org/10.18653/v1/2026.acl-long.2152}
}

References

38 references, 15 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Yirong Chen, Xiaofen Xing, Jingkai Lin, Huimin Zheng, Zhenyu Wang, Qi Liu, and Xiangmin Xu. 2023. Soulchat: Improving llms empathy, listening, and comfort abilities through fine-tuning with multi-turn empathy conversations. In Findings of the Association for Computational Linguistics: EMNLP 2023, pa… find
  2. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, and 1 others. 2025. Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. arX…
  3. Carolyn E Cutrona and Daniel W Russell. 1990. Type of social support and specific stress: Toward a theory of optimal matching. find
  4. Faiza Farhat. 2024. Chatgpt as a complementary mental health resource: a boon or a bane. Annals of Biomedical Engineering, 52(5):1111–1114. find
  5. John O Greene and Brant R Burleson. 2003. Handbook of communication and social interaction skills. Routledge. find
  6. Jiahe Guo, Xiangran Guo, Yulin Hu, Zimo Long, Xingyu Sui, Xuda Zhi, Yongbo Huang, Hao He, Weixiang Zhao, Yanyan Zhao, and 1 others. 2026. When personalization legitimizes risks: Uncovering safety vulnerabilities in personalized dialogue agents. arXiv preprint arXiv:2601.17887. find
  7. Han Han, Tong Zhu, Xiang Zhang, MengSong Wu, Xiong Hao, and Wenliang Chen. 2025. NesTools: A dataset for evaluating nested tool learning abilities of large language models. In Proceedings of the 31st International Conference on Computational Linguistics, pages 9824–9844, Abu Dhabi, UAE. Association…
  8. Catherine A Heaney and Barbara A Israel. 2008. Social networks and social support. Health behavior and health education: Theory, research, and practice, 4(1):189–210. find
  9. Edward J Hu, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, and 1 others. 2022. Lora: Low-rank adaptation of large language models. In International Conference on Learning Representations.
  10. Yulin Hu, Zimo Long, Jiahe Guo, Xingyu Sui, Xing Fu, Weixiang Zhao, Yanyan Zhao, and Bing Qin. 2026. Op-bench: Benchmarking over-personalization for memory-augmented personalized conversational agents. arXiv preprint arXiv:2601.13722. — [arXiv:2601.13722](https://arxiv.org/abs/2601.13722) find
  11. Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, and 1 others. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276.
  12. Bowen Jiang, Zhuoqun Hao, Young-Min Cho, Bryan Li, Yuan Yuan, Sihao Chen, Lyle Ungar, Camillo J Taylor, and Dan Roth. 2025. Know me, respond to me: Benchmarking llms for dynamic user profiling and personalized responses at scale. arXiv preprint arXiv:2504.14225. — [arXiv:2504.14225](https://arxiv.org/abs/2504.14225) find
  13. Kim Klyver, Benson Honig, and Paul Steffens. 2017. Social support timing and persistence in nascent entrepreneurship: exploring when instrumental and emotional support is most effective. Small Business Economics. find
  14. Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph Gonzalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serving with pagedattention. In Proceedings of the 29th symposium on operating systems principles, pages 611–626.
  15. Catherine Penny Hinson Langford, Juanita Bowsher, Joseph P Maloney, and Patricia P Lillis. 1997. Social support: a conceptual analysis. Journal of advanced nursing, 25(1):95–100. find
  16. Minghao Li, Yingxiu Zhao, Bowen Yu, Feifan Song, Hangyu Li, Haiyang Yu, Zhoujun Li, Fei Huang, and Yongbin Li. 2023. API-bank: A comprehensive benchmark for tool-augmented LLMs. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 3102–3116, Singapore. Ass…
  17. June M Liu, Donghao Li, He Cao, Tianhe Ren, Zeyi Liao, and Jiamin Wu. 2023. Chatcounselor: A large language models for mental health support. arXiv preprint arXiv:2309.15461. — [arXiv:2309.15461](https://arxiv.org/abs/2309.15461) find
  18. Siyang Liu, Chujie Zheng, Orianna Demasi, Sahand Sabour, Yu Li, Zhou Yu, Yong Jiang, and Minlie Huang. 2021. Towards emotional support dialog systems. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Lan…
  19. Xiao Liu, Hao Yu, Hanchen Zhang, Yifan Xu, Xuanyu Lei, Hanyu Lai, Yu Gu, Hangliang Ding, Kaiwen Men, Kejuan Yang, Shudan Zhang, Xiang Deng, Aohan Zeng, Zhengxiao Du, Chenhui Zhang, Sheng Shen, Tianjun Zhang, Yu Su, Huan Sun, and 3 others. 2025. Agentbench: Evaluating llms as agents. Preprint, arXiv:…
  20. Adyasha Maharana, Dong-Ho Lee, Sergey Tulyakov, Mohit Bansal, Francesco Barbieri, and Yuwei Fang. 2024. Evaluating very long-term conversational memory of llm agents. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 13851–13870… find
  21. Grégoire Mialon, Roberto Dessì, Maria Lomeli, Christoforos Nalmpantis, Ram Pasunuru, Roberta Raileanu, Baptiste Rozière, Timo Schick, Jane Dwivedi-Yu, Asli Celikyilmaz, Edouard Grave, Yann LeCun, and Thomas Scialom. 2023. Augmented language models: a survey. Preprint, arXiv:2302.07842.
  22. OpenAI. 2025. Introducing GPT-4.1 in the API. Blog post. Accessed: 2025-12-30. find
  23. Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, and 1 others. 2019. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32.
  24. Wei Peng, Yue Hu, Luxi Xing, Yuqiang Xie, Yajing Sun, and Yunpeng Li. 2022. Control globally, understand locally: A global-to-local hierarchical graph network for emotional support conversation. arXiv preprint arXiv:2204.12749. — [arXiv:2204.12749](https://arxiv.org/abs/2204.12749) find
  25. Yujia Qin, Shengding Hu, Yankai Lin, Weize Chen, Ning Ding, Ganqu Cui, Zheni Zeng, Xuanhe Zhou, Yufei Huang, Chaojun Xiao, Chi Han, Yi Ren Fung, Yusheng Su, Huadong Wang, Cheng Qian, Runchu Tian, Kunlun Zhu, Shihao Liang, Xingyu Shen, and 24 others. 2024. Tool learning with foundation models. ACM Co…
  26. Huachuan Qiu, Hongliang He, Shuai Zhang, Anqi Li, and Zhenzhong Lan. 2023. Smile: Single-turn to multi-turn inclusive language expansion via chatgpt for mental health support. arXiv preprint arXiv:2305.00450.
  27. Norbert K Semmer, Achim Elfering, Nicola Jacobsha-gen, Tanja Perrot, Terry A Beehr, and Norbert Boos. 2008. The emotional meaning of instrumental social support. International journal of stress management, 15(3):235. find
  28. Omar Shaikh, Kristina Gligoric, Ashna Khetan, Matthias Gerstgrasser, Diyi Yang, and Dan Jurafsky. 2024. Grounding gaps in language model generations. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Vol… find
  29. Harsh Trivedi, Tushar Khot, Mareike Hartmann, Ruskin Manku, Vinty Dong, Edward Li, Shashank Gupta, Ashish Sabharwal, and Niranjan Balasubramanian. 2024. AppWorld: A controllable world of apps and people for benchmarking interactive coding agents. In Proceedings of the 62nd Annual Meeting of the Asso…
  30. Quan Tu, Yanran Li, Jianwei Cui, Bin Wang, Ji-Rong Wen, and Rui Yan. 2022. Misc: A mixed strategy-aware model integrating comet for emotional support conversation. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 308– 319.
  31. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, and 1 others. 2025. Qwen3 technical report. arXiv preprint arXiv:2505.09388.
  32. Yuxiang Zhang, Jing Chen, Junjie Wang, Yaxin Liu, Cheng Yang, Chufan Shi, Xinyu Zhu, Zihao Lin, Hanwen Wan, Yujiu Yang, Tetsuya Sakai, Tian Feng, and Hayato Yamana. 2024. ToolBeHonest: A multi-level hallucination diagnostic benchmark for tool-augmented large language models. In Proceedings of the 20… find
  33. Haiquan Zhao, Lingyu Li, Shisong Chen, Shuqi Kong, Jiaan Wang, Kexin Huang, Tianle Gu, Yixu Wang, Jian Wang, Liang Dandan, and 1 others. 2024. Esc-eval: Evaluating emotion support conversations in large language models. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language P… find
  34. Weixiang Zhao, Xingyu Sui, Xinyang Han, Yang Deng, Yulin Hu, Jiahe Guo, Libo Qin, Qianyun Du, Shijin Wang, Yanyan Zhao, and 1 others. 2025. Chain of strategy optimization makes large language models better emotional supporter. arXiv preprint arXiv:2503.05362. find
  35. Weixiang Zhao, Yanyan Zhao, and Xin Lu. 2022. Cauain: Causal aware interaction network for emotion recognition in conversations. In IJCAI, pages 4524–4530. find
  36. Weixiang Zhao, Yanyan Zhao, Xin Lu, Shilong Wang, Yanpeng Tong, and Bing Qin. 2023a. Is chatgpt equipped with emotional dialogue capabilities? arXiv preprint arXiv:2304.09582. — [arXiv:2304.09582](https://arxiv.org/abs/2304.09582) find
  37. Weixiang Zhao, Yanyan Zhao, Shilong Wang, and Bing Qin. 2023b. Transesc: Smoothing emotional support conversation via turn-level state transition. In Findings of the Association for Computational Linguistics: ACL 2023, pages 6725–6739. find
  38. Zhonghua Zheng, Lizi Liao, Yang Deng, Libo Qin, and Liqiang Nie. 2024. Self-chats from large language models make small emotional support chatbot better. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 11325–11345. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/sui-2026-tea-bench-systematic-benchmarking-tool/