FLUX.1 Krea [dev] is a 12B text-to-image model from Black Forest Labs and Krea. It came out 31 July 2025 under terms that allow non-commercial use. It’s a fixed-up flow transformer tuned for nice-looking photos, solid prompt-following, and quick results. The model’s open for researchers and creative folks.
They trained it using guidance distillation so it samples faster than regular diffusion models. It’s open but still tries to match closed models in image quality and prompt handling. You can plug it into setups like ComfyUI or HuggingFace Diffusers in place of FLUX.1 [dev].
It delivers sharp visuals, strong prompt response, and tools that fit creative use. You can use it personally or in science stuff, and even for money-making work if it sticks to the Flux-1-dev-Non-Commercial rules. Still, you can’t use it for shady stuff like breaking laws, fake content, or hurting people.
They added safety checks before and after training. That includes:
Filtered out NSFW and illegal content before training.
Fine-tuned to block fake CSAM and forced content.
Worked with Internet Watch Foundation and other testers to check how tough it is.
Posted filters and tracking tools on GitHub.
BFL can review how it’s used and ban folks who break the terms.
If you'd like to access this model, you can explore the following possibilities:
Use our video cost calculator to compare prices between platforms offering FLUX.1 Krea [dev] model.
For locally hosted models, see description and additional links at the bottom for versions, repos and tutorials.
