<code>FLUX.1 [schnell]</code> is a 12 billion parameter rectified flow transformer capable of generating images from text descriptions. It is trained using latent adversarial diffusion distillation, allowing it to generate high-quality images in only 1 to 4 steps. The model offers cutting-edge output quality and competitive prompt following, matching the performance of closed-source alternatives. Released under the Apache 2.0 license, the model can be used for personal, scientific, and commercial purposes. You must provide proper attribution, indicate if changes were made, and not use the model or its derivatives for unlawful or harmful activities. Follow official ComfyUI installation, workflow and usage: <a target="_blank" rel="ugc" href="https://comfyanonymous.github.io/ComfyUI_examples/flux/">https://comfyanonymous.github.io/ComfyUI_examples/flux/</a>Key Features:<ul><li>High-quality image generation from text descriptions.</li></ul>Efficient image generation in 1 to 4 steps. Out-of-Scope UseThe model and its derivatives may not be used<ul><li>In any way that violates any applicable national, federal, state, local or international law or regulation.</li><li>For the purpose of exploiting, harming or attempting to exploit or harm minors in any way; including but not limited to the solicitation, creation, acquisition, or dissemination of child exploitative content.</li><li>To generate or disseminate verifiably false information and/or content with the purpose of harming others.</li><li>To generate or disseminate personal identifiable information that can be used to harm an individual.</li><li>To harass, abuse, threaten, stalk, or bully individuals or groups of individuals.</li><li>To create non-consensual nudity or illegal pornographic content.</li><li>For fully automated decision making that adversely impacts an individual's legal rights or otherwise creates or modifies a binding, enforceable obligation.</li><li>Generating or facilitating large-scale disinformation campaigns. </li></ul>Usage:<ul><li>Reference implementation and sampling code are available in a dedicated Black Forest Labs <a target="_new" rel="ugc" href="https://github.com/black-forest-labs/flux">GitHub repository</a>.</li></ul>Credits: Developed by Black Forest Labs. FLUXby Black Forest Labs: <a target="_blank" rel="ugc" href="https://blackforestlabs.ai">https://blackforestlabs.ai</a>

FLUX.1

float8_e4m3fn weights of:<a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.1-schnell">https://huggingface.co/black-forest-labs/FLUX.1-schnell</a>Licenseflux1-schnell-fp8.safetensors falls under the <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/choosealicense/licenses/blob/main/markdown/apache-2.0.md">Apache-2.0 License</a>.Version credit to: <a target="_blank" rel="ugc" href="https://huggingface.co/Kijai">Kijai</a>

flux1_schnellFP8Kijai11GB.safetensors

Schnell FP8 Kijai [11 GB]

DEV FP8 - Kijai [11 GB]

Schnell

flux1-schnell-fp8 is a derivative of FLUX.1, the latest image generation AI model. This model has the following features:## Main FeaturesHigh-quality image generationBased on the FLUX.1 [schnell] model, it can generate high-quality images quickly.LightweightQuantized in FP8 (8-bit floating point) format, which significantly reduces memory usage compared to the original FP16 model.Use on low-spec PCsDue to the lightweight design, it can run on relatively low-spec PCs and GPUs. It has been reported that it can be used in environments with 8GB VRAM + 16GB RAM.## Technical Features- Uses T5+CLIP as a text encoder- Uses a unique DiT (Diffusion Transformer) for the dediffusion part- Model distilled from FLUX.1 [pro] using Guidance Distillation## NotesAs a trade-off for the lightweight nature of the FP8 format, there may be a slight degradation in image quality and generation speed compared to the FP16 version. However, the quality remains sufficient for many users.flux1-schnell-fp8 is a groundbreaking model that combines high-quality image generation capabilities with low resource requirements, and is used by a wide range of users.

flux1-schnell-fp8

**flux1-schnell-fp8** is a derivative of **FLUX.1**, the latest image generation AI model. This model has the following features:

## Main Features

**High-quality image generation**
Based on the FLUX.1 [schnell] model, it can generate high-quality images quickly.

**Lightweight**
Quantized in FP8 (8-bit floating point) format, which significantly reduces memory usage compared to the original FP16 model.

**Use on low-spec PCs**
Due to the lightweight design, it can run on relatively low-spec PCs and GPUs. It has been reported that it can be used in environments with 8GB VRAM + 16GB RAM.

## Technical Features

- Uses T5+CLIP as a text encoder
- Uses a unique DiT (Diffusion Transformer) for the dediffusion part
- Model distilled from FLUX.1 [pro] using Guidance Distillation

## Notes

As a trade-off for the lightweight nature of the FP8 format, there may be a slight degradation in image quality and generation speed compared to the FP16 version. However, the quality remains sufficient for many users.

**flux1-schnell-fp8** is a groundbreaking model that combines high-quality image generation capabilities with low resource requirements, and is used by a wide range of users.

Files with SHA256: bbdfba27fed8ff3be237523fb37b83821a6c4bbaa1db43ef9288767d0e4042fb