This model card is aimed for those who wish to run SDXL models in GGUF quantized format.

Due to the steps required to extract UNet and Clip Encoders and convert the UNet model to a GGUF quantized model, I decided to share the models already quantized in GGUF format here. Out of the vast number of SDXL finetuned models, only a few models are currently quantized mostly for my personal use. I will continue to add more as time allows.

Since each SDXL finetuned model has its own licensing term, you should check the repo of the original finetuned model for licensing information as the quantized version will fall under the same license. The most of the finetuned models can be found on CivitAI.

If you are using a v_prediction model, the quantized model won't be automatically detected as such in ComfyUI. I made a custom node to force the model type to set as v_prediction in ComfyUI for the purpose. V_Prediction Node Link: https://github.com/magekinnarus/ComfyUI-V-Prediction-Node

If you are the owner of any finetuned model featured and do not wish for your model to be available in GGUF quantized format, please contact me at: https://www.reddit.com/user/OldFisherman8/