A 4B-parameter world-action model for physical AI. Given an initial scene image and a task instruction, Kairos predicts the future video showing the robot performing the described action.
📄 Paper · 💻 GitHub · 🤗 Model