Black Forest Labs opens early access to FLUX 3
Black Forest Labs has opened early access to the video model in FLUX 3, a single system trained on images, video and audio together, and says early tests show users prefer it to most rivals.
Black Forest Labs announced FLUX 3, a multimodal foundation model trained on images, video and audio at the same time in one architecture. Only FLUX 3 Video is available for now, in early access. The company says early access for FLUX 3 Image will open "in the following weeks."
The model builds on Self-Flow, the company's method for handling multimodal generation and understanding in the same architecture. Black Forest Labs says it trained FLUX 3 with much more compute and data than before. The video model can produce clips of up to 20 seconds with native audio in a single generation. Supported modes include text-to-video, image-to-video, video-to-video, keyframe-to-video, continuing an existing video and its audio, and multilingual dialogue. Users can also chain clips into multi-shot sequences several minutes long.
The company tested 10-second, 720p text-to-video clips with audio. It reports that FLUX 3 was preferred over Luma Ray 3.2 in 93% of comparisons, Runway Gen-4.5 in 77%, Grok Imagine Video in up to 69% and Kling v3 Pro in 60%. Against Seedance 2.0 and Gemini Omni Flash, the figure was 52%, which is close to even. Black Forest Labs calls these results preliminary. It ran the evaluations itself and did not describe how they were conducted.
The company is also pitching the model for robotics. With mimic robotics, it built FLUX-mimic, a model that pairs FLUX 3's video backbone with action prediction for dexterous manipulation. The two companies are testing it on production tasks at Audi.
The source text is cut off in the launch-plan section, so pricing, API details and the full release schedule are not yet known.