Daily AI CatchupBlack Forest Labs Opens Access to FLUX 3
Black Forest Labs opened early access to FLUX 3, a "visual intelligence" model that generates 20-second video clips with synced multilingual audio from text, image, or video inputs — beating Runway, Kling, and Grok Imagine on output quality in BFL's own testing. Its companion FLUX-mimic variant, built with Zurich robotics startup mimic robotics, trains factory robots on Audi's production line from roughly 30 minutes of demonstration footage instead of the usual 30+ hours.
Why it matters
💻 Developer · The 30-minutes-vs-30-hours reduction in demo data for FLUX-mimic is the number worth tracking if you're anywhere near robotics or imitation-learning work.
📦 Product · A single model spanning content generation and physical robot training suggests BFL is positioning as infrastructure across both markets — worth watching if either is on your roadmap.
🎨 Design · 20-second audio-synced clips claimed to beat Runway/Kling/Grok Imagine deserves a hands-on look once wider access opens, especially for motion-graphics or ad-creative workflows.
📈 Business · Pairing a closed early-access release with a planned open-weight version aimed at factory hardware suggests BFL is chasing enterprise licensing and open-source mindshare at the same time.
🤔 Just Curious · A new AI model from Black Forest Labs can make realistic 20-second videos with sound, and a special version of it can teach factory robots new tasks by watching just 30 minutes of a person doing it, instead of the usual 30+ hours.
Sources: Black Forest Labs: FLUX 3