Black Forest Labs launched FLUX 3 in early access, its first model that generates video up to 20 seconds with synchronized audio, moving beyond the still-image generation that built FLUX’s reputation. The model was preferred over Runway Gen-4.5 in 77% of head-to-head comparisons and over Luma Ray 3.2 in 93% of cases. FLUX-mimic, a robotics variant developed with mimic robotics, translates the model’s video predictions into physical robot motions; Audi is already testing this technology on soft-body door seal fitting tasks on its production lines. The system reacts in approximately 101 milliseconds, comparable to human visual reflexes. Access is currently restricted to APIs and select partners, with the open-weight Dev version not planned until late 2026.
Source: Read the original article
