FLUX 3 Action predicts the next move and the next state of the world from camera feeds alone.
The multimodal model generates images, 20-second video clips with sound, and even predicts physical actions from a single…