Article

RynnUnify: 16B unified embodied model combines physical understanding with long-horizon visual planning

The authors propose RynnUnify, a 16B unified embodied model that combines embodied understanding with long-horizon visual planning through video-language interleaved reasoning. The model covers spatial reasoning, physical prediction, and grounding capabilities, linking physical world understanding with visual anticipation across subtasks. The post also provides links to the code and paper.

Part of the event: RynnUnify Releases 16B Unified Embodied Model

This page provides a summary and source link only.