Black Forest Labs launches FLUX 3 Video with Crusoe AI infrastructure support

Black Forest Labs has entered the video generation race with FLUX 3, its first multimodal foundation model. The company announced the model on July 23, 2026, with Crusoe AI working through the weekend to provide the underlying hardware support needed to get the launch off the ground at scale. FLUX 3 supports text-to-video and image-to-video creation, generates clips up to 20 seconds in HD quality, and produces native audio alongside the visual output. Early access to the model became available through the BFL API and to a set of selected partners in late July 2026. A dedicated blog post focused on the video generation capabilities followed on August 4, 2026, and Crusoe AI’s Anjney Midha publicly highlighted the team’s intensive efforts to meet the compute demands of the rollout on August 6, 2026. The model’s applications extend beyond consumer media creation. Black Forest Labs has been building toward robotics use cases, with existing collaborations involving mimic robotics and Audi already in place. FLUX 3’s multimodal architecture is designed to feed into those pipelines, and the company has indicated plans to release an open-weight version of the FLUX 3 Dev backbone at a future ...