Black Forest Labs releases FLUX 3, a multimodal base model.

Black Forest Labs has launched FLUX 3, a multimodal foundational model that, for the first time, jointly learns images, videos, and audio within a unified architecture. Based on Self-Flow technology, the model can generate up to 20 seconds of video with native audio in a single pass, supporting text-generated and image-generated video modes. By learning world representations through intermodal physical constraints, it demonstrates excellent performance in robot manipulation tasks.