FLUX.1

FLUX.1 is the image generation model family from Black Forest Labs, a company founded by former Stability AI researchers. The models use a rectified flow transformer architecture and come in two open-weight text-to-image variants: schnell, a fast few-step model released under Apache 2.0, and dev, a guidance-distilled model under a non-commercial license. Specialized editions extend the base capability to inpainting and outpainting (Fill), structural conditioning from Canny edges or depth maps, image variation (Redux), and instruction-based editing (Kontext), with LoRA versions of the conditioning models available. The official repository ships minimal PyTorch inference code requiring Python 3.10 or newer, with optional TensorRT acceleration, and the same models are served through a documented REST API. Running locally typically calls for a GPU with 12 GB or more of VRAM. Developers, artists, and researchers use FLUX.1 both directly and through front ends such as ComfyUI and Diffusers.