12B parameter rectified flow transformer for photorealism, typography, and prompt adherence.
FLUX is an open and commercial suite of 12-billion parameter flow-matching models created by the original Stable Diffusion inventors at Black Forest Labs. It sets the open benchmark for visual anatomy, text rendering, and cinematic optical prompt execution.
Official model variants, context limits, and target filmmaking workflows
State-of-the-art enterprise image generation API with extreme prompt fidelity and nuanced cinematic grain texture.
Open-weights non-commercial model for local pipeline experimentation and fine-tuning (LoRA / ControlNet).
Verified analysis and production case studies
A rigorous peer comparison evaluating FLUX.1 [dev] and Midjourney v6.1 across high-contrast chiaroscuro key lighting, skin subsurface scattering, and anamorphic lens bokeh.
Relationships mapped to canonical tools, techniques, prompts, workflows, films, and research
Every factual capability, benchmark, and parameter is backed by a verifiable primary source
Black Forest Labs GmbH