Black Forest Labs releases FLUX 3 Action, an open-weights robotics model claiming a leaderboard-topping result at half the parameter count of rivals.

Black Forest Labs releases FLUX 3 Action, an open-weights robotics model claiming a leaderboard-topping result at half the parameter count of rivals.

Black Forest Labs released FLUX 3 Action, an open-weights AI model for robotics that reportedly tops its benchmark leaderboard while using roughly half the parameters of competing models. The release extends the FLUX family, previously known for image and video generation, into physical-world action and robotics tasks.

The move puts a generative-media foundation-model company directly into the robotics segment at a moment when general-purpose foundation models are increasingly being tested as a substitute for task-specific robot control stacks. Choosing an open-weights release for a robotics model signals that Black Forest Labs is betting on distribution and adoption in a still-forming benchmark landscape rather than proprietary lock-in, and the claimed efficiency, a top leaderboard result at roughly half the parameter count of competitors, pushes back on the assumption that stronger robotics performance requires larger models.

For builders, an open-weights robotics model from an established image and video lab lowers the barrier to experimenting with foundation-model-driven robot control without training from scratch or depending on closed APIs. For investors, Black Forest Labs, which has raised $465M in total funding per the AI Market Watch index (coverage tracks roughly 5,000 companies, not a census), is signaling ambition beyond its core generative-media business, a diversification worth watching as other image and video labs test whether their architectures generalize to physical-world tasks.

#BlackForestLabs #FLUX3Action #Robotics #OpenWeights #FoundationModels #PhysicalAI