Black Forest Labs Unveils FLUX 3 AI: Ditches Stills for Video—And Robot Hands
Black Forest Labs launched FLUX 3, its first flagship model to generate video as well as images. It is trained jointly on images, video, and audio in one multimodal system, and can produce clips up to 20 seconds with synchronized dialogue, effects, and ambient sound. In early human preference tests, FLUX 3 was chosen over Runway Gen-4.5 in 77% of comparisons and Luma Ray 3.2 in 93%, with a slight edge over Gemini Omni and Seedance. It also remains strong at still-image generation across many styles. BFL says video prediction may help models learn physical dynamics, not just content creation. That idea powers FLUX-mimic, a robotics system built with mimic robotics that adds a decoder to turn video understanding into robot motion; Audi is testing it on flexible assembly tasks. FLUX 3 is in early access via APIs and partners now, with broader image access soon and an open-weight dev release planned for later in 2026.
