Breaking: Black Forest Labs, the team behind the FLUX image models, just stepped into robotics.
They released FLUX 3 Action, a 7B parameter open-weights World Action Model, and it is already topping the RoboLab-120 benchmark.
Here is the simple version of what that means.
Most robots today just react. You give a command, they move, and that is it. No sense of what happens next.
FLUX 3 Action does something smarter. It watches the camera feed, checks the robot's current state, reads your text instruction, then does two things at once.
It imagines the future video frames and plans the next chunk of actions, simultaneously.
Think of it like a chess player who visualizes the board three moves ahead before touching a single piece.
That is basically what happens before a robotic arm even twitches.
And because it is open-weights, this is not locked inside a lab at Google or Tesla. Any researcher or startup can download it and start building.
Robots just got a little closer to thinking before they act.
🔗 Original Source & Reference: https://www.marktechpost.com/2026/09/24/black-forest-labs-releases-flux-3-action-a-7b-open-weights-world-action-model-that-tops-robolab-120/
Published automatically via FeedMind AI Content Pipeline.
Top comments (0)