Elamur Research⌕
☰
home / papers / 2026 / AgenticQwen: Training Small Agentic Language Models with Dual Data Flywheels for Industrial-Scale Tool Use

AgenticQwen: Training Small Agentic Language Models with Dual Data Flywheels for Industrial-Scale Tool Use

Paper · 2026 · Yuanjie Lyu, Chengyu Wang, Haonan Zheng et al.

Summary

Industrial agent systems often run on large proprietary or open models, which costs too much at scale. Major labs rarely release small models with strong tool-use skills. This Alibaba paper presents AgenticQwen, a family of compact agents that use small Qwen models as their foundation. Their training relies on GRPO-style multi-round reinforcement learning, drawing mostly on synthetic data plus some open-source data. Training has two parts:

  • Reasoning RL: math and search problems using web-search and code tools, rewarded only on final-answer correctness.
  • Agentic RL: simulated users and tools, scored by rubric-based subgoal checks.

Single-round RL plateaus because synthetic data is homogeneous, so the authors add two "data flywheels":

  • Reasoning flywheel: turns the model's failures into harder, verified math variants.
  • Agentic flywheel: grows simple linear workflows into multi-branch behavior trees, with simulated users sometimes trying to mislead the agent.

After three flywheel rounds, the 8B and 30B-A3B models far exceed their vanilla Qwen3 versions on agentic benchmarks and approach Qwen3-235B. In a production analytics/search system, they also infer faster. The work shows that cheap small models can handle routine, high-frequency tool-use tasks.

Key points

  • Benchmark averages: AgenticQwen-8B scores 47.4 versus 23.8 for vanilla Qwen3-8B. AgenticQwen-30B-A3B, an MoE model with about 3B active parameters, scores best overall at 50.2. Gains rise steadily from Round 0 to Round 3 across seven task categories.
  • Seed data: Omni, 2WikiMultiHopQA and HotpotQA for reasoning RL; SynthAgent for agentic RL. Qwen3-235B simulates the users and tools.
  • Hard-example generation: new problems come from Self-Instruct rewriting and persona injection. A sample is kept only if Qwen3-235B reaches the same answer in three independent solutions.
  • Industrial search benchmarks: with under 10K agentic search training examples, AgenticQwen-30B-A3B beats vanilla Qwen3-30B-A3B by +17.0 on XBench. Mean GAIA run time is 344.1 s, versus 355.6 s for vanilla Qwen3-30B-A3B and 449.5 s for Qwen3-235B.
  • Released resources: checkpoints and part of the synthetic data are on Hugging Face (alibaba-pai/agenticqwen), and code is on GitHub. The data pipeline is also integrated into EasyDistill.
  • Limitations: the 40k context limit hurts deep-search tasks, and using Qwen models throughout the pipeline may introduce model-family bias.

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

🏢Alibaba GroupHangzhou · China168 works · 15,386 citations

 Fields, topics and concepts

📚 Shelves 4The 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 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 3,320 works

Cite

Cite as Yuanjie Lyu, Chengyu Wang, Haonan Zheng, Yuanhao Yue, Junbing Yan, Ming Wang and Jun Huang. 2026. AgenticQwen: Training Small Agentic Language Models with Dual Data Flywheels for Industrial-Scale Tool Use. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-industry.37

@inproceedings{lyu2026agenticqwen,
  title = {{AgenticQwen: Training Small Agentic Language Models with Dual Data Flywheels for Industrial-Scale Tool Use}},
  author = {Yuanjie Lyu, Chengyu Wang, Haonan Zheng, Yuanhao Yue, Junbing Yan, Ming Wang and Jun Huang},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-industry.37},
  url = {https://doi.org/10.18653/v1/2026.acl-industry.37}
}

References

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

  1. for medical question-answering tasks. Int. J. Medical Informatics, 212:106339. find
  2. Anthropic. 2025. Claude. https://claude.ai. ↗
  3. Yifan Bai, Yiping Bao, Guanduo Chen, Jiahao Chen, Ningxin Chen, Ruijue Chen, Yanru Chen, Yuankun Chen, Yutian Chen, Zhuofu Chen, Jialei Cui, Hao Ding, Mengnan Dong, Angang Du, Chenzhuang Du, Dikang Du, Yulun Du, Yu Fan, Yichen Feng, and 80 others. 2025. Kimi k2: Open agentic intelligence. arXiv prep…
  4. Victor Barres, Honghua Dong, Soham Ray, Xujie Si, and Karthik Narasimhan. 2025. τ 2 -bench: Evaluating conversational agents in a dual-control environment. arXiv preprint arXiv:2506.07982. find
  5. Wenrui Cai, Chengyu Wang, Junbing Yan, Jun Huang, and Xiangzhong Fang. 2025. Enhancing reasoning abilities of small llms with cognitive alignment. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 7434–7449. find
  6. Hunter Lightman, Vineet Kosaraju, Yuri Burda, Harrison Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. 2023. Let’s verify step by step. In The Twelfth International Conference on Learning Representations.
  7. Yuanjie Lyu, Chengyu Wang, Jun Huang, and Tong Xu. 2025. From correction to mastery: Reinforced distillation of large language model agents. arXiv preprint arXiv:2509.14257. — [arXiv:2509.14257](https://arxiv.org/abs/2509.14257) find
  8. Yuanjie Lyu, Chengyu Wang, Lei Shen, Jun Huang, and Tong Xu. 2026. Mock worlds, real skills: Building small agentic language models with synthetic tasks, simulated environments, and rubric-based rewards. arXiv preprint arXiv:2601.22511.
  9. Aili Chen, Aonian Li, Bangwei Gong, Binyang Jiang, Bo Fei, Bo Yang, Boji Shan, Changqing Yu, Chao Wang, Cheng Zhu, Chengjun Xiao, Chengyu Du, Chi Zhang, Chu Qiao, Chunhao Zhang, Chunhui Du, Congchao Guo, Da Chen, Deming Ding, and 80 others. 2025a. Minimax-m1: Scaling test-time compute efficiently wi…
  10. Kaiyuan Chen, Yixin Ren, Yang Liu, Xiaobo Hu, Haotong Tian, Tianbao Xie, Fangfu Liu, Haoye Zhang, Hongzhang Liu, Yuan Gong, and 1 others. 2025b. xbench: Tracking agents productivity scaling with profession-aligned real-world evaluations. arXiv preprint arXiv:2506.13651. — [arXiv:2506.13651](https://arxiv.org/abs/2506.13651) find
  11. DeepSeek-AI. 2025. Deepseek-v3.2: Pushing the frontier of open large language models. arXiv preprint arXiv:2512.02556.
  12. Bofei Gao, Feifan Song, Zhe Yang, Zefan Cai, Yibo Miao, Qingxiu Dong, Lei Li, Chenghao Ma, Liang Chen, Runxin Xu, and 1 others. 2025. Omni-math: A universal olympiad level mathematic benchmark for large language models. In The Thirteenth International Conference on Learning Representations.
  13. Tao Ge, Xin Chan, Xiaoyang Wang, Dian Yu, Haitao Mi, and Dong Yu. 2025. Scaling synthetic data creation with 1,000,000,000 personas. arXiv preprint arXiv:2406.20094.
  14. Xanh Ho, Anh-Khoa Duong Nguyen, Saku Sugawara, and Akiko Aizawa. 2020. Constructing a multi-hop qa dataset for comprehensive evaluation of reasoning steps. In Proceedings of the 28th International Conference on Computational Linguistics, pages 6609– 6625.
  15. Shuyue Jia, Subhrangshu Bit, Varuna H. Jasodanand, Yi Liu, and Vijaya B. Kolachalama. 2026. Agentic memory-augmented retrieval and evidence grounding Weizhen Li, Jianbo Lin, Zhuosong Jiang, Jingyi Cao, Xinpeng Liu, Jiayu Zhang, Zhenqiang Huang, Qianben Chen, Weichen Sun, Qiexiang Wang, Hongxuan Lu,… — [arXiv:2508.13167](https://arxiv.org/abs/2508.13167) find
  16. Grégoire Mialon, Clémentine Fourrier, Thomas Wolf, Yann LeCun, and Thomas Scialom. 2023. Gaia: a benchmark for general ai assistants. In The Twelfth International Conference on Learning Representations.
  17. OpenAI. 2025. Gpt-5. https://openai.com/ gpt-5. ↗
  18. Shishir G. Patil, Huanzhi Mao, Charlie Cheng-Jie Ji, Fanjia Yan, Vishnu Suresh, Ion Stoica, and Joseph E. Gonzalez. 2025. The berkeley function calling leaderboard (bfcl): From tool use to agentic evaluation of large language models. In Forty-second International Conference on Machine Learning. find
  19. Zeyu Qin, Qingxiu Dong, Xingxing Zhang, Li Dong, Xiaolong Huang, Ziyi Yang, Mahmoud Khademi, Dongdong Zhang, Hany Hassan Awadalla, Yi R. Fung, Weizhu Chen, Minhao Cheng, and Furu Wei. 2025. Scaling laws of synthetic data for language models. arXiv preprint arXiv:2503.19551. — [arXiv:2503.19551](https://arxiv.org/abs/2503.19551) find
  20. Partha Pratim Ray. 2025. A review on vibe coding: Fundamentals, state-of-the-art, challenges and future directions. Authorea Preprints. find
  21. John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. 2017. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347. find
  22. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Mingchuan Zhang, Y. K. Li, Y. Wu, and Daya Guo. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300.
  23. Minjie Shen, Yanshu Li, Lulu Chen, and Qikai Yang. 2025. From mind to machine: The rise of manus ai as a fully autonomous digital agent. arXiv preprint arXiv:2505.02024. — [arXiv:2505.02024](https://arxiv.org/abs/2505.02024) find
  24. Yi Su, Dian Yu, Linfeng Song, Juntao Li, Haitao Mi, Zhaopeng Tu, Min Zhang, and Dong Yu. 2025. Crossing the reward bridge: Expanding rl with verifiable rewards across diverse domains. arXiv preprint arXiv:2503.23829. — [arXiv:2503.23829](https://arxiv.org/abs/2503.23829) find
  25. Li Sun, Liu He, Shuyue Jia, Yangfan He, and Chenyu You. 2025. Docagent: An agentic framework for multi-modal long-context document understanding. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, EMNLP 2025, Suzhou, China, November 4-9, 2025, pages 17701–1771… find
  26. Chengyu Wang, Junbing Yan, Wenrui Cai, Yuanhao Yue, and Jun Huang. 2025. Easydistill: A comprehensive toolkit for effective knowledge distillation of large language models. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 787–795… find
  27. Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A Smith, Daniel Khashabi, and Hannaneh Hajishirzi. 2023. Self-instruct: Aligning language models with self-generated instructions. In Proceedings of the 61st annual meeting of the association for computational linguistics (volume 1: long p… find
  28. Jialong Wu, Wenbiao Yin, Yong Jiang, Zhenglin Wang, Zekun Xi, Runnan Fang, Linhai Zhang, Yulan He, Deyu Zhou, Pengjun Xie, and 1 others. 2025. Webwalker: Benchmarking llms in web traversal. arXiv preprint arXiv:2501.07572.
  29. Zhiheng Xi, Wenxiang Chen, Xin Guo, Wei He, Yiwen Ding, Boyang Hong, Ming Zhang, Junzhe Wang, Senjie Jin, Enyu Zhou, Rui Zheng, Xiaoran Fan, Xiao Wang, Limao Xiong, Yuhao Zhou, Weiran Wang, Changhao Jiang, Yicheng Zou, Xiangyang Liu, and 9 others. 2025. The rise and potential of large language model…
  30. Xiaohan Xu, Ming Li, Chongyang Tao, Tao Shen, Reynold Cheng, Jinyang Li, Can Xu, Dacheng Tao, and Tianyi Zhou. 2024. A survey on knowledge distillation of large language models. arXiv preprint arXiv:2402.13116. — [arXiv:2402.13116](https://arxiv.org/abs/2402.13116) find
  31. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, and 40 others. 2025. Qwen3 technical report. arXiv preprint arXiv:2505.09388.
  32. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. 2018. HotpotQA: A dataset for diverse, explainable multi-hop question answering. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing.
  33. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. 2022. React: Synergizing reasoning and acting in language models. In The eleventh international conference on learning representations.
  34. Asaf Yehudai, Lilach Eden, Alan Li, Guy Uziel, Yilun Zhao, Roy Bar-Haim, Arman Cohan, and Michal Shmueli-Scheuer. 2025. Survey on evaluation of llm-based agents. arXiv preprint arXiv:2503.16416. — [arXiv:2503.16416](https://arxiv.org/abs/2503.16416) find
  35. Guibin Zhang, Hejia Geng, Xiaohang Yu, Zhenfei Yin, Zaibin Zhang, Zelin Tan, Heng Zhou, Zhongzhi Li, Xiangyuan Xue, Yijiang Li, Yifan Zhou, Yang Chen, Chen Zhang, Yutao Fan, Zihu Wang, Songtao Huang, Yue Liao, Hongru Wang, Mengyue Yang, and 6 others. 2025. The landscape of agentic reinforcement lear… — [arXiv:2509.02547](https://arxiv.org/abs/2509.02547) find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/lyu-2026-agenticqwen-training-small-agentic-language/