FLUX 3 Turns an Image Model Into an Image, Video, Audio and Robot Action Model

Black Forest Labs announced FLUX 3 on 23 July 2026, a multimodal frontier model that learns jointly from images, video and audio in a single unified architecture, extensible to action prediction.

FLUX 3 Video

Generates clips up to 20 seconds in a single generation, with native audio. Supported modes: text-to-video, image-to-video, video-to-video from a reference clip, keyframe-to-video for controlled transitions, and generative video-audio continuation from input video and audio.

FLUX-mimic

An action model developed with Swiss robotics startup Mimic Robotics that carries out instructions - Black Forest Labs' move into physical AI. It is being tested with manufacturing partners including Audi.

What you can actually use today

This is where expectations need managing. Video and Action are in early access only, via API and private weights to selected partners. FLUX 3 Image was described at announcement as arriving "in the coming weeks". The open-weight FLUX 3 Dev release is planned for later in 2026.

Roster note: our roster carries FLUX.2 Pro, which remains the current generally-available FLUX model. We will not move the roster to FLUX 3 until the Image release is broadly available and its specifications are documented - announcing a model is not the same as shipping it, and a partner-only early access programme is not something readers can evaluate or buy.