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

• URL: http://arxiv.org/abs/2511.11793v2
• Date: Tue, 18 Nov 2025 15:45:29 GMT
• Title: MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling
• Authors: MiroMind Team, Song Bai, Lidong Bing, Carson Chen, Guanzheng Chen, Yuntao Chen, Zhe Chen, Ziyi Chen, Jifeng Dai, Xuan Dong, Wenhan Dou, 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,
• Abstract summary: MiroThinker is an open-source research agent designed to advance tool-augmented reasoning and information-seeking capabilities.<n>Unlike previous agents that only scale up model size or context length, MiroThinker explores interaction scaling at the model level.
• Score: 115.74855199827596
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We present MiroThinker v1.0, an open-source research agent designed to advance tool-augmented reasoning and information-seeking capabilities. Unlike previous agents that only scale up model size or context length, MiroThinker explores interaction scaling at the model level, systematically training the model to handle deeper and more frequent agent-environment interactions as a third dimension of performance improvement. Unlike LLM test-time scaling, which operates in isolation and risks degradation with longer reasoning chains, interactive scaling leverages environment feedback and external information acquisition to correct errors and refine trajectories. Through reinforcement learning, the model achieves efficient interaction scaling: with a 256K context window, it can perform up to 600 tool calls per task, enabling sustained multi-turn reasoning and complex real-world research workflows. Across four representative benchmarks-GAIA, HLE, BrowseComp, and BrowseComp-ZH-the 72B variant achieves up to 81.9%, 37.7%, 47.1%, and 55.6% accuracy respectively, surpassing previous open-source agents and approaching commercial counterparts such as GPT-5-high. Our analysis reveals that MiroThinker benefits from interactive scaling consistently: research performance improves predictably as the model engages in deeper and more frequent agent-environment interactions, demonstrating that interaction depth exhibits scaling behaviors analogous to model size and context length. These findings establish interaction scaling as a third critical dimension for building next-generation open research agents, complementing model capacity and context windows.

Related papers

• ASTER: Agentic Scaling with Tool-integrated Extended Reasoning [27.877412657068806]
  Reinforcement learning (RL) has emerged as a dominant paradigm for eliciting long-horizon reasoning in Large Language Models (LLMs)<n>We introduce ASTER (Agentic Scaling with Tool-integrated Extended Reasoning), a framework that circumvents this collapse through a targeted cold-start strategy.<n>We find that a small expert cold-start set of just 4K interaction-dense trajectories yields the strongest downstream performance.
  arXiv  Detail & Related papers  (2026-02-01T12:46:02Z)
• WebSeer: Training Deeper Search Agents through Reinforcement Learning with Self-Reflection [51.10348385624784]
  We present WebSeer, a more intelligent search agent trained via reinforcement learning enhanced with a self-reflection mechanism.<n>Our approach substantially extends tool-use chains and improves answer accuracy.
  arXiv  Detail & Related papers  (2025-10-21T16:52:00Z)
• Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window [88.85901839023803]
  DeepMiner is a novel framework that elicits such abilities by introducing high-difficulty training tasks and dynamic context window.<n>We develop DeepMiner-32B, which achieves substantial performance improvements across multiple search agent benchmarks.
  arXiv  Detail & Related papers  (2025-10-09T14:31:39Z)
• OmniEAR: Benchmarking Agent Reasoning in Embodied Tasks [52.87238755666243]
  We present OmniEAR, a framework for evaluating how language models reason about physical interactions, tool usage, and multi-agent coordination in embodied tasks.<n>We model continuous physical properties and complex spatial relationships across 1,500 scenarios spanning household and industrial domains.<n>Our systematic evaluation reveals severe performance degradation when models must reason from constraints.
  arXiv  Detail & Related papers  (2025-08-07T17:54:15Z)
• Thinking vs. Doing: Agents that Reason by Scaling Test-Time Interaction [46.286440953594266]
  We propose to scale test-time interaction, an untapped dimension of test-time scaling.<n>We first show that even prompting-based interaction scaling can improve task success on web benchmarks non-trivially.<n>We introduce TTI (Test-Time Interaction), a curriculum-based online reinforcement learning approach that trains agents by adaptively adjusting their rollout lengths.
  arXiv  Detail & Related papers  (2025-06-09T17:50:02Z)
• Thinking Longer, Not Larger: Enhancing Software Engineering Agents via Scaling Test-Time Compute [61.00662702026523]
  We propose a unified Test-Time Compute scaling framework that leverages increased inference-time instead of larger models.<n>Our framework incorporates two complementary strategies: internal TTC and external TTC.<n>We demonstrate our textbf32B model achieves a 46% issue resolution rate, surpassing significantly larger models such as DeepSeek R1 671B and OpenAI o1.
  arXiv  Detail & Related papers  (2025-03-31T07:31:32Z)
• Can foundation models actively gather information in interactive environments to test hypotheses? [43.42688356541211]
  Foundation models excel at single-turn reasoning but struggle with multi-turn exploration in dynamic environments.<n>We evaluated these models on their ability to learn from experience, adapt, and gather information.
  arXiv  Detail & Related papers  (2024-12-09T12:27:21Z)
• Tool-Augmented Reward Modeling [58.381678612409]
  We propose a tool-augmented preference modeling approach, named Themis, to address limitations by empowering RMs with access to external environments. Our study delves into the integration of external tools into RMs, enabling them to interact with diverse external sources. In human evaluations, RLHF trained with Themis attains an average win rate of 32% when compared to baselines.
  arXiv  Detail & Related papers  (2023-10-02T09:47:40Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.