ABENZERPS / UNCENSORED QUANTIZED

Turn an idea into
a reproducible render.

Compose prompts and sampling settings for the exact GGUF and Safetensors files in the model repository. Copy one command, or export a compact recipe for ComfyUI.

Try hosted live demo ↗Open exact model ↗
01 / ComposeLOCAL RECIPE
02 / Render recipe

        

        
Exact weightsDiffusion + Qwen3-VL encoder + Qwen Image 2.1 VAE, all from the linked repository.
Memory postureKeep the diffusion model on GPU; offload the text encoder to system RAM on smaller cards.
Safety noteThis release has no built-in safety checker. Follow the laws and policies that apply where you run it.

      

    

    
03 / Workflow notes
Built for the files
you actually downloaded.
A
ComfyUI native
Use Unet Loader (GGUF), a Qwen Image text encoder, and the matching VAE. The repo's README includes the official workflow templates.
Read setup ↗
B
stable-diffusion.cpp
The generated command follows the Qwen Image 2.1 sd.cpp recipe: Euler sampling, flow shift 3, and dimensions divisible by 32.
Read recipe ↗
C
Hosted generation
Need an output now? Open a hosted Qwen Image 2.1 demo. It is a separate hosted pipeline, not a claim that this browser-only page runs the quantized weights.
Open hosted demo ↗