Black Forest Labs releases FLUX 3 Action, an open robotics model
Black Forest Labs has launched FLUX 3 Action, an open AI model specifically designed for robotics applications. Building on the multimodal FLUX 3 foundation, which was trained on video, image, and audio data, this new model functions as a world-action model. It processes multi-camera video feeds from a robot's workspace to predict the next action an agent should take and how the environment will…
Key points
- FLUX 3 Action uses seven billion parameters, less than half the size of the previous best open model.
- The model runs up to 3.95 times faster than the previous best open model on the RoboLab-120 leaderboard.
- Weights are available on Hugging Face, allowing for on-device deployment and digital environment testing.
According to Black Forest Labs, FLUX 3 Action achieves a record success rate on the RoboLab-120 leaderboard. The model uses only seven billion parameters, making it less than half the size of the previous best open model. It also runs up to 3.95 times faster than its predecessor. The company argues that while large reasoning models can plan effectively, they are often too slow and bulky for real-time robot deployment, making this efficiency critical for on-device use.
Beyond physical robotics, Black Forest Labs sees potential for the model in digital environments. Video games could serve as testing grounds for navigation and fast-reacting agents that might eventually operate computers. The model weights are available for download on Hugging Face.
Model page: FLUX 3 Action →
Black Forest Labs launches FLUX 3 Action, an open robotics AI model
The Decoder · 24 September 2026
Loading the full article…
This text was published by The Decoder and written by Matthias Bastian. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Robotics & Physical AI
All →- Global industrial robot stock hits 5 million in 2025, IFR says · 1 src
- Mexican team builds portable robotics-AI lab for schools · 1 src
- Epson launches AX6 cobot with 6 kg payload and no-code programming · 1 src
- Shield AI, Waabi, and GM discuss deploying AI where failure isn’t an option · 1 src
- State of Humanoids keynote features Agility, Apptronik, Persona AI, Psyonic at RoboBusiness · 1 src
Comments
via GitHub Discussions