nunchaku-flux.1-dev-colossus Open-source Image Generation Model - Produce high-quality images based on text prompts with efficient inference

Nunchaku Flux.1 Dev Colossus

Developed by nunchaku-tech

The Nunchaku quantized version of the Colossus Project Flux, designed to generate high-quality images based on text prompts. This model minimizes performance loss while optimizing inference efficiency.

Image Generation EnglishOpen Source License:Other #4-bit Quantized Text-to-Image Generation #SVDQuant Low-Rank Compression #Non-Commercial License

Downloads 235

Release Time : 7/23/2025

Model Overview

This is a text-to-image generation model, quantized based on the FLUX.1-dev model, suitable for generating high-quality images.

Model Features

Efficient Quantization

Adopt SVDQuant quantization technology to compress the model into INT4 or NVFP4 format, significantly improving inference efficiency.

High Performance

Minimize performance loss while optimizing inference efficiency, maintaining high-quality image generation capabilities.

Multi-Hardware Support

Provide quantized versions suitable for different GPU architectures, including non-Blackwell GPUs and Blackwell GPUs.

Model Capabilities

Text-to-Image Generation

High-Quality Image Generation

Efficient Inference

Use Cases

Creative Design

Artistic Creation

Generate artworks based on text prompts.

High-quality art images

Concept Design

Quickly generate design concept diagrams.

Diverse design sketches

Content Generation

Social Media Content

Generate attractive social media images.

Highly appealing visual content

🚀 Model Card for nunchaku-flux.1-dev-colossus

This repository offers Nunchaku - quantized versions of Colossus Project Flux, enabling high - quality image generation from text prompts. It's optimized for efficient inference with minimal performance loss.

Nunchaku Logo

visual

✨ Features

Generate high - quality images from text prompts.
Optimized for efficient inference with minimal performance loss.

📦 Installation

No installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

Diffusers Usage: Refer to flux.1-dev-colossus.py. Simply replace the safetensors with the ones from this repository. For more advanced usage, check our tutorial.
ComfyUI Usage: Refer to nunchaku-flux.1-dev.json. Just replace the safetensors with the ones from this repository.

📚 Documentation

Model Details

Model Description

Property	Details
Developed by	Nunchaku Team
Model Type	text - to - image
License	flux-1-dev-non-commercial-license
Quantized from model	FLUX.1-dev

Model Files

: SVDQuant quantized INT4 FLUX ColossusV12 model. For users with non - Blackwell GPUs (pre - 50 - series).
: SVDQuant quantized NVFP4 FLUX ColossusV12 model. For users with Blackwell GPUs (50 - series).

Model Sources

Inference Engine: nunchaku
Quantization Library: deepcompressor
Paper: SVDQuant: Absorbing Outliers by Low - Rank Components for 4 - Bit Diffusion Models
Demo: svdquant.mit.edu

Performance

performance

Citation

@inproceedings{
  li2024svdquant,
  title={SVDQuant: Absorbing Outliers by Low - Rank Components for 4 - Bit Diffusion Models},
  author={Li*, Muyang and Lin*, Yujun and Zhang*, Zhekai and Cai, Tianle and Li, Xiuyu and Guo, Junxian and Xie, Enze and Meng, Chenlin and Zhu, Jun - Yan and Han, Song},
  booktitle={The Thirteenth International Conference on Learning Representations},
  year={2025}
}

Attribution Notice

The FLUX.1 [dev] Model is licensed by Black Forest Labs Inc. under the FLUX.1 [dev] Non - Commercial License. Copyright Black Forest Labs Inc. IN NO EVENT SHALL BLACK FOREST LABS INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.