MEGA Hub

RynnUnify: Towards Unified Embodied Brain

Authors

Do you know Minghao Zhu?You can claim authorship or link another user.Do you know Bohan Hou?You can claim authorship or link another user.Do you know Kun Xiang?You can claim authorship or link another user.Do you know Zixuan Wang?You can claim authorship or link another user.Do you know Yinzhou Tang?You can claim authorship or link another user.Do you know Yifei Yang?You can claim authorship or link another user.Do you know Zhikai Wang?You can claim authorship or link another user.Do you know Tianyi Zhang?You can claim authorship or link another user.Do you know Kehan Li?You can claim authorship or link another user.Do you know Haoyu Zhao?You can claim authorship or link another user.Do you know Siteng Huang?You can claim authorship or link another user.Do you know Leigang Qu?You can claim authorship or link another user.Do you know Sicong Leng?You can claim authorship or link another user.Do you know Dongchi Huang?You can claim authorship or link another user.Do you know Mingxiu Chen?You can claim authorship or link another user.Do you know Shijian Lu?You can claim authorship or link another user.Do you know Yue Wang?You can claim authorship or link another user.Do you know Xiaodan Liang?You can claim authorship or link another user.Do you know Zhenyu Zeng?You can claim authorship or link another user.Do you know Deli Zhao?You can claim authorship or link another user.Do you know Chengju Liu?You can claim authorship or link another user.Do you know Jianfei Yang?You can claim authorship or link another user.Do you know Xin Li?You can claim authorship or link another user.

Abstract

We introduce RynnUnify, a 16B unified embodied model that combines strong embodied understanding with long-horizon visual planning through interleaved video--language reasoning. With capabilities spanning spatial reasoning, physical prediction, and grounding, RynnUnify connects an understanding of the physical world with visual anticipation of how it evolves across successive subtasks. Building on these capabilities, we further develop an agentic extension through reinforcement learning to selectively use video predictions for reasoning about unobserved states and interaction outcomes. We also develop a world-action extension of the full RynnUnify model, translating its understanding and prediction capabilities into robot control. To evaluate these capabilities, we introduce Rynn-VP and Rynn-GEU for long-horizon visual planning and anticipatory reasoning. Experiments show that RynnUnify achieves the highest average embodied understanding score among the evaluated unified multimodal models and improves the Rynn-VP overall score by 9.75 points over Cosmos 3 Nano. Its agentic extension improves the Rynn-GEU average by 6.0 points over direct reasoning, while the world-action extension achieves 51.5\% success on 20 selected composite RoboCasa365 tasks and 46.7\% average success across three real-robot tasks. Together, these results establish RynnUnify as a shared foundation for embodied understanding, visual planning, and robot control, taking a step toward a unified embodied brain.

Community

00