State-of-the-Art Speed
Designed for efficiency, FLUX.1 [schnell] is the fastest in the FLUX.1 family, enabling rapid iteration and real-time applications.
FLUX.1 [schnell] is a 12 billion parameter rectified flow transformer capable of generating high-quality images from text descriptions. It is optimized for speed and efficiency, making it perfect for personal use and local integration.
The perfect balance of speed and performance for developers and creators.
Designed for efficiency, FLUX.1 [schnell] is the fastest in the FLUX.1 family, enabling rapid iteration and real-time applications.
Built on a hybrid architecture of multimodal and parallel diffusion transformer blocks, scaled to 12B parameters for superior performance.
Openly available for personal use and local development. Experiment, build, and integrate freely with an open-source license.
Maintains high visual quality and prompt adherence, surpassing many strong non-distilled models in its class.
Supports a wide range of aspect ratios and resolutions, preserving the diversity seen in pretraining data.
Available on HuggingFace, with inference code on GitHub and Diffusers, plus day-1 ComfyUI support.
Create stunning images in seconds with rectified flow technology.
Describe your vision in natural language. FLUX.1 [schnell] excels at prompt adherence, understanding complex descriptions with ease.
Adjust aspect ratio and settings. Despite its speed, FLUX.1 [schnell] offers control over your image generation process.
Experience lightning-fast generation. The distilled model architecture ensures rapid results without compromising on core visual quality.
Get high-resolution outputs ready for your personal projects or local development needs under the Apache 2.0 license.
FLUX.1 [schnell] FAQ
FLUX.1 [schnell] is the fastest text-to-image model in the FLUX.1 suite by Black Forest Labs. It is a 12 billion parameter rectified flow transformer tailored for local development and personal use.
Yes, FLUX.1 [schnell] is released under the Apache 2.0 license, making it openly available for personal use and local development.
FLUX.1 [schnell] is the distilled, efficiency-focused version, faster than [dev] and [pro]. While [pro] offers the absolute peak performance, [schnell] provides the best trade-off for speed and local usability.
The model is released under Apache 2.0. Please refer to the specific license terms on the HuggingFace repository for commercial usage details and restrictions.
As a 12B parameter model, it requires decent GPU memory. However, it supports 1-4 step inference and can be quantized (e.g., FP8) for better efficiency on consumer hardware.
Yes, like other FLUX.1 models, it has strong capabilities in typography and prompt following, allowing for accurate text rendering in generated images.
Weights are available on HuggingFace under 'black-forest-labs/FLUX.1-schnell'.
Yes, it has day-1 integration with ComfyUI and is supported by the HuggingFace Diffusers library.
Rectified flow is a generative training method that connects data and noise in a straight line, improving efficiency compared to standard diffusion models.
Absolutely. It preserves the photorealistic capabilities of the FLUX.1 family, generating highly detailed and realistic images.
The perfect balance of speed and performance for developers and creators.
Try It Now