Black Forest Labs — the team behind the FLUX image models — has announced FLUX 3: a single multimodal model that generates and understands image, video, and audio together, with early research work on action-prediction for robotics as well.
We get asked a lot whether GenMedia supports FLUX 3 yet. Short answer: not yet, and we'd rather say that plainly than add a button that doesn't work.
What's actually available right now
As of this writing, FLUX 3 is in what Black Forest Labs themselves call "Early Access" — not a public, generally-available API. It isn't part of our current FLUX lineup yet either — today Canvas and Apps support FLUX.2 Pro, FLUX.2 Klein, FLUX.2 Turbo, FLUX Pro Kontext, FLUX.1 Dev, and FLUX LoRA.
In practice that means there is no real, callable endpoint for us to wire up yet — anyone offering a working "FLUX 3" generator today is either using early-access access we don't have, or the button doesn't actually call FLUX 3.
What we did instead
Rather than stay silent about it, we added a "FLUX 3 (Coming soon)" entry to Canvas's node palette. It's visible, but deliberately not draggable onto the canvas — we'd rather show you it's on our radar than ship something that fails every time you click it.
As soon as FLUX 3 has a real, generally-available API we can call, we'll wire it into Canvas and the Apps catalog the same way as every other model — no separate migration, no waitlist on our end.
What FLUX 3 is expected to bring
Per Black Forest Labs' own announcement, FLUX 3 is built to generate image, video, and audio jointly from one model, with improved prompt-following and multilingual text rendering over earlier FLUX versions. We'll publish an update here the moment it's something you can actually run.