16 upvotes
RynnUnify: 16B Model Merges Vision, Language for Real-World Planning
What is RynnUnify: Towards Unified Embodied Brain?
RynnUnify is a 16-billion-parameter model that watches video, understands language, and reasons about physical actions over long horizons. It can predict object motion, solve spatial puzzles, and ground textual commands in visual scenes.
Who it is for
Robotics and embodied-AI teams can use it to prototype agents that plan and act in complex visual environments.
How to try it
Visit the model page on Hugging Face to download the weights.