Black Forest Labs, the German AI studio based in Freiburg, announced FLUX 3 on July 23, its first video model and its first attempt to bundle several types of media generation into one system.
Unlike tools that stitch separate models together, FLUX 3 was trained jointly on images, video, audio and even robot actions using an approach the company calls Self-Flow. Its headline trick is generating clips of up to 20 seconds with sound produced in the same pass, so the picture and audio come out together rather than being matched afterwards.
The model is being released across four lines: FLUX 3 Video, FLUX 3 Image, FLUX 3 Action and an open-weight version called FLUX 3 Dev that is due later in 2026. As of July 24, only Video and Action were reachable, and only through gated early access that the company approves case by case. No public pricing or open API had been announced.
Chief executive Robin Rombach explained the thinking behind training on many signals at once. “You can’t cheat reality. A model that only learns images can only generate images. That’s why FLUX 3 is trained across those signals together, so the model can build a deeper understanding of how the world works,” he said.
In the company’s own tests, run on early 10-second clips, FLUX 3 was preferred 93 percent of the time over Luma Ray 3.2 and 77 percent over Runway Gen-4.5, while roughly tying with Google’s Gemini Omni Flash and Seedance 2.0. A robotics version built with Zurich firm mimic is already being trialled by carmaker Audi on factory tasks, with a stated response time of about 101 milliseconds. Design partners testing the tools include Canva, Krea, Picsart and Magnific.
Black Forest Labs, founded in August 2024 by researchers from the original Stable Diffusion team, has raised more than 450 million dollars and was valued at 3.25 billion dollars in December 2025.
Why it matters: tools that turn a few typed words into short videos with sound could dramatically lower the cost of making adverts, music visuals and social content for African creators.
Comments 0