← All model drops
Live drop
FLUX 3 Action
Black Forest LabsDropped Sep 23, 20263 builds3 makers1 prompt
Open-weight 7B world-action model that jointly predicts robot motor commands and future video frames from camera + state + text.
Live drop
Black Forest LabsDropped Sep 23, 20263 builds3 makers1 prompt
Open-weight 7B world-action model that jointly predicts robot motor commands and future video frames from camera + state + text.
About this release
FLUX 3 Action is a 7B-parameter open-weight multimodal model (FLUX Kommunity License) built on the FLUX 3 backbone. It ingests camera frames, robot state, and a text instruction to output the next action chunk plus predicted video of the scene. Includes base checkpoint, DROID and SO-101 policies, LeRobot integration, and training code on GitHub.
Read the official Black Forest Labs launch pageWorth opening first
Paste the URL and Dropday.ai will credit you, track the build as it evolves, and put it on this drop.
Submit a build