Black Forest Labs makes FLUX 3 Video generally available: 20-second clips with native audio

Black Forest Labs, the German lab founded by the original Stable Diffusion authors, has opened its first video model to everyone. FLUX 3 Video, announced in late July in early access, became generally available on August 5 through the company's API and select partners.

The distinctive trait is that video comes out already sounding right: instead of generating frames first and bolting on audio with a second tool, the model produces clips up to 20 seconds with synchronized dialogue, sound effects, and ambient noise in a single pass. Lip-sync works in more than 14 languages: you write the line and the character speaks it with coherent mouth movement.

For content creators the workflow covers the usual cases: start from text or an image, pin keyframes, continue an existing video, or switch scenes and camera angles inside the same clip. Pricing is per second of generated video: from 6 US cents per second in draft mode up to roughly 53 cents for Full HD video-to-video, audio always included.

The usual caveat applies: comparisons against competitors (ByteDance's Seedance 2.0 above all) come from the lab's internal evaluations, not independent tests. But with open weights and 2K/4K resolutions announced as coming, the move puts pressure on the sector's giants in a territory, video with native audio, that until a few months ago belonged to two or three names.