CoolFace
Modelpublic

npario/Qwen-Image-Edit-mflux

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face
0likes960downloads
Model Card

Qwen Image Edit 2511 — Pre-quantized for mflux

Pre-quantized weights for Qwen/Qwen-Image-Edit-2511, ready to use with mflux on Apple Silicon.

Available quantizations

VariantFolderSize
3-bitQwen-Image-Edit-2511-q3~23 GB
4-bitQwen-Image-Edit-2511-q4~25 GB
5-bitQwen-Image-Edit-2511-q5~28 GB
6-bitQwen-Image-Edit-2511-q6~30 GB
8-bitQwen-Image-Edit-2511-q8~35 GB

Full (unquantized) weights: ~54 GB from the original repo.

Usage

Download a specific quantization:

bash
huggingface-cli download fcreait/Qwen-Image-Edit-mflux Qwen-Image-Edit-2511-q4 --local-dir ./qwen-edit-q4

Generate with mflux:

bash
mflux-generate-qwen-edit \
  --model ./qwen-edit-q4/Qwen-Image-Edit-2511-q4 \
  --base-model qwen \
  --image-paths input.jpg \
  --prompt "your editing prompt" \
  --output output.png

Details

  • Quantized with mflux v0.17.4 using mflux-save
  • Source model: Qwen/Qwen-Image-Edit-2511
  • Format: MLX safetensors with mflux metadata
  • Includes VisionTransformer weights (required for image editing)

License

These weights are quantized derivatives of Qwen/Qwen-Image-Edit-2511, released by Alibaba Cloud under the Apache License 2.0. Quantization does not change the license — these derivatives are distributed under the same terms.

Copyright the Qwen team, Alibaba Cloud. Licensed under the Apache License, Version 2.0. A copy of the license is available at <http://www.apache.org/licenses/LICENSE-2.0>.

Quantization by CREAIT. mflux is MIT licensed.