Skip to content

Source paper

MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling

MiroMind Team, Song Bai, Lidong Bing and 51 more

MiroMind Team, Song Bai, Lidong Bing, Carson Chen, Guanzheng Chen, Yuntao Chen, Zhe Chen, Ziyi Chen, Jifeng Dai, Xuan Dong, Yue Deng, Yunjie Fu, Junqi Ge, Chenxia Han, Tammy Huang, Zhenhang Huang, Jerry Jiao, Shilei Jiang, Tianyu Jiao, Xiaoqi Jian, Lei Lei, Ruilin Li, Ryan Luo, Tiantong Li, Xiang Lin, Ziyuan Liu, Zhiqi Li, Jie Ni, Qiang Ren, Pax Sun, Shiqian Su, Chenxin Tao, Bin Wang, Hellen Wang, Haonan Wang, James Wang, Jin Wang, Jojo Wang, Letian Wang, Shizun Wang, Weizhi Wang, Zixuan Wang, Jinfan Xu, Sen Xing, Chenyu Yang, Hai Ye, Jiaheng Yu, Yue Yu, Muyan Zhong, Tianchen Zhao, Xizhou Zhu, Yanpeng Zhou, Yifan Zhang, Zhi Zhu

arXiv 2511.11793 · 2025165 upvotes on Hugging Face

Research ideas from this paper (3)

Generated by AI. The paper’s authors did not write these.

AI summary

MiroThinker v1.0 is an open-source research agent designed to push performance boundaries through 'interactive scaling,' where the model is trained to handle deeper and more frequent agent-environment interactions. By utilizing a 256K context window and a three-stage training pipeline involving SFT, DPO, and Reinforcement Learning (GRPO), the agent achieves state-of-the-art results on benchmarks like GAIA and BrowseComp, rivaling proprietary models like GPT-5-high.

How it works (AI summary)

The system is built on Qwen2.5/3 models using a ReAct paradigm. The training pipeline consists of Agentic Supervised Fine-Tuning (SFT) on synthetic data, Agentic Preference Optimization (DPO) to align decision-making, and Agentic Reinforcement Learning using Group Relative Policy Optimization (GRPO) to encourage exploration and error correction. Context is managed via a Recency-Based Context Retention mechanism that masks older tool outputs to save tokens.