Black Forest Labs unveils FLUX 3, a unified multimodal frontier model

Black Forest Labs unveiled FLUX 3 on 23 July 2026, a new multimodal frontier model that jointly learns from images, video and audio within a single unified architecture and can be extended to predict actions.

What's new: Where earlier FLUX releases focused on image generation and editing, FLUX 3 positions the German startup as building a "foundation layer for visual intelligence" — one network spanning image, video, audio and action prediction. Video with native audio generation is now in early access, with further capabilities to be rolled out over the coming weeks and months after phased early-access testing.

Into physical AI: The company also introduced FLUX-mimic, an action model developed with Swiss startup Mimic Robotics that can carry out instructions, marking a shift towards physical AI. It is being tested with manufacturing partners including carmaker Audi.

Read more: Black Forest Labs · Bloomberg