Why FLUX 3 Trains on Images, Video and Sound — Andreas Blattmann & Dustin Podell, Black Forest Labs
modelsLatent SpaceNov 16, 2026
What can a model learn about the world by watching video and listening to it? Black Forest Labs trained FLUX 3 on images, video and audio together. Andreas Blattmann and Dustin Podell join swyx to explain why. A moving camera reveals the far side of an...
Impact: LowConfidence: LowNovelty: Incremental
modelslatent-spacemultimodal