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

See the whole story

Read the original on Latent Space

Sens.ai aggregates and assesses published reporting. The assessment above is machine generated; the original source is authoritative.