Black Forest Labs launches FLUX 3 for image and video generation

Freiburg, Germany-based AI lab Black Forest Labs (BFL) has expanded its FLUX family with the launch of FLUX 3. The new model is capable of understanding and generating both images and video clips up to 20 seconds long, complete with audio, from a single text prompt.

BFL emphasizes that FLUX 3 was trained jointly across these modalities, rather than stitching together separate image, video, and audio models. The company pitches this as a unified capability called "visual intelligence," applicable to enterprise needs in creative generation, simulation, and robotics.

FLUX 3 Video and FLUX 3 Action products are now entering a limited "Early Access" program, for which any party can apply. The company states that FLUX 3 Image will roll out in the coming weeks, followed by broader general availability. Specific pricing and service-level agreements have not yet been announced.

In preliminary comparisons, BFL claims FLUX 3 outperformed competitors. In head-to-head tests involving 10-second video clips with audio, FLUX 3 reportedly achieved a 93% preference rate over Luma Ray 3.2. BFL notes that these are preliminary evaluations and full benchmark results will be published later.