Black Forest Labs launches Flux 3: a model that creates images, videos, and audio together

Black Forest Labs has released Flux 3, a new model that can handle images, videos, and audio all at once. It uses a single system to learn from all three types of content, building on the earlier Flux 1 and Flux 2 versions.

The model can generate videos up to 20 seconds long with built-in sound. It supports creating video from text or images, extending existing clips, turning keyframes into video, and multi-language chats. In tests with 10-second, 720p videos with sound, Flux 3 beat Grok Imagine Video 69% of the time, and tied with Seedance 2.0 and Gemini Omni Flash at 52% win rates.

Black Forest Labs is working with Mimic Robotics to use Flux 3 for predicting robot behavior. For images, the model can create and edit content in various styles, sizes, and resolutions.