FLUX (image model) FLUX.1 / Schnell / Pro
More info
- Creator
- Black Forest Labs (ex-Stable Diffusion team)
- Released
- Type
- Text-to-image models
- Website
- blackforestlabs.ai
Image-model family from the ex-Stability founders (Robin Rombach et al.) — Schnell (open), Dev (non-commercial), Pro (managed API).
Rectified flow transformer (different from traditional diffusion). Much better text rendering than SDXL, fingers usually work, prompt adherence on par with DALL·E 3. The Schnell variant is optimised for 4-step generation. Quickly became the new standard for open-weight images. FLUX.1 Kontext (2024) adds image editing via prompts.