Robotics · 2026-10-26

RynnUnify: Towards Unified Embodied Brain

Minghao Zhu, Bohan Hou, Kun Xiang, Zixuan Wang, Yinzhou Tang, Yifei Yang, Zhikai Wang, Tianyi Zhang, Kehan Li, Haoyu Zhao, Siteng Huang, Leigang Qu, Sicong Leng, Dongchi Huang, Mingxiu Chen, Shijian Lu, Yue Wang, Xiaodan Liang, Zhenyu Zeng, Deli Zhao, Chengju Liu, Jianfei Yang, Xin Li

arXiv:2610.32965PDF

AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.

TL;DR

RynnUnify is a 16B unified embodied model (MoT-based) that combines an autoregressive reasoner with a diffusion rectified-flow video generator via interleaved video–language reasoning. It is evaluated on long-horizon visual planning (Rynn-VP, Rynn-GEU) and extended to agentic reasoning (RL) and robot control (WAM).

Why it matters

The authors aim to unify embodied understanding, long-horizon visual planning, and robot control by connecting understanding-guided reasoning with temporally coherent video prediction. They introduce Rynn-VP and Rynn-GEU to evaluate sustained visual planning and anticipatory reasoning about unobserved future states, and report improvements for the base model plus agentic and world-action extensions.

Method

  • RynnUnify uses an autoregressive reasoner (retained from Cosmos3) plus a rectified-flow diffusion video generator in an MoT architecture; the branches interact through joint attention.
  • Interleaved video–language reasoning: the reasoner predicts the next executable subtask and a detailed description of anticipated visual evolution and duration, which conditions the generator; generated clip outcomes feed subsequent reasoning.
  • Extensions: an agentic version post-trained with reinforcement learning to selectively invoke frozen RynnUnify for video generation (vs direct answering), and a World-Action Model (WAM) via latent planning and joint video-action training for robot control.
Abstract (from arXiv)

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.

Related papers

Publication record

Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.

First public release: 2026-10-23 · Indexed by Pub-AI: 2026-10-26 UTC

preprint

RynnUnify: Towards Unified Embodied Brain

Minghao Zhu, Bohan Hou, Kun Xiang, Zixuan Wang, Yinzhou Tang, Yifei Yang, Zhikai Wang, Tianyi Zhang, Kehan Li, Haoyu Zhao, Siteng Huang, Leigang Qu, Sicong Leng, Dongchi Huang, Mingxiu Chen, Shijian Lu, Yue Wang, Xiaodan Liang, Zhenyu Zeng, Deli Zhao, Chengju Liu, Jianfei Yang, Xin Li

Preprint posted: 2026-10-23 · Original record ↗

arxiv
2610.32965
Explicit source versions (1)
  • v1 · 2026-10-23
  1. Preprint available

    Source date: 2026-10-23 · First detected: 2026-10-27 UTC · Historical baseline

    Source evidence (2)
Abstract · arxiv

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.