<h1 id="qwen-image-2.1-ggufsgguf:-https:huggingface.corealrebelaiqwen-image-2.1_ggufstreemain">Qwen-Image 2.1 GGUFs GGUF: <a target="_blank" rel="ugc" href="https://huggingface.co/realrebelai/Qwen-Image-2.1_GGUFs/tree/main">https://huggingface.co/realrebelai/Qwen-Image-2.1_GGUFs/tree/main</a> </h1>GGUF quantizations of Qwen-Image 2.1 for use in ComfyUI with ComfyUI-GGUF.These files are converted from the Comfy-Org Qwen-Image 2.1 BF16 diffusion model and preserve the native Comfy tensor layout. No additional tensor-name remapping is required after conversion.<blockquote>This repository contains the diffusion model only. You still need the normal Qwen-Image 2.1 supporting models required by your ComfyUI workflow, such as the text encoder and VAE.</blockquote><h2 id="quantizations">Quantizations</h2>Planned / available quant ladder:QuantRelative SizeNotesQ8_0Largest quantHighest precision quantized optionQ6_KLargeVery high quality / good precisionQ5_K_MMedium-LargeHigh qualityQ4_K_MMediumRecommended balance of size and qualityQ3_K_MSmallLower VRAM / more aggressive quantizationQ2_KSmallestMost aggressive quantization; quality may degradeQuantization quality can vary by prompt, resolution, sampler, and workflow. Test lower-bit versions before relying on them for production work.<h2 id="source">Source</h2>BF16 diffusion model source:Comfy-Org/Qwen-Image-2.1<code>diffusion_models/qwen_image_2.1_bf16.safetensors</code>Official / upstream Qwen-Image 2.1 model components and configuration are from the Qwen project.<h2 id="architecture">Architecture</h2>The GGUF files use:<pre><code>general.architecture = qwen_image
</code></pre>The source checkpoint contains:<ul><li>32 transformer blocks</li><li>4096 hidden dimension</li><li>32 attention heads</li><li>128 head dimension</li><li>64 input channels</li><li>64 output channels</li></ul>The Comfy-Org BF16 checkpoint already uses the tensor naming/layout expected by ComfyUI, so the conversion preserves those names instead of remapping them to a separate llama.cpp-style naming scheme.<h2 id="precision-handling">Precision handling</h2>The repeating transformer blocks are quantized according to the selected GGUF format.Small or sensitive non-repeating tensors are preserved at higher precision where appropriate, including components such as:<pre><code>img_in
txt_in
time_text_embed
modulation
norm_out
proj_out
</code></pre>The conversion also stores <code>comfy.gguf.orig_shape.*</code> metadata where a tensor must be physically reshaped for GGUF quantization compatibility. ComfyUI-GGUF restores the logical tensor shape when loading.<h2 id="comfyui-installation">ComfyUI Installation</h2>Install or update ComfyUI-GGUF:<pre><code>cd /d D:\AI_Tools\ComfyUI_windows_portable\ComfyUI\custom_nodes
git clone https://github.com/city96/ComfyUI-GGUF
</code></pre>Then install its requirements with ComfyUI's embedded Python:<pre><code>D:\AI_Tools\ComfyUI_windows_portable\python_embeded\python.exe -m pip install -r D:\AI_Tools\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-GGUF\requirements.txt
</code></pre>If you already have ComfyUI-GGUF installed, update it instead of cloning a second copy.<h2 id="model-placement">Model Placement</h2>Place the GGUF diffusion model in your normal ComfyUI diffusion-model directory, for example:<pre><code>ComfyUI/models/diffusion_models/
</code></pre>Then load it with:<pre><code>Unet Loader (GGUF)
</code></pre>or the equivalent GGUF diffusion-model loader supplied by your installed ComfyUI-GGUF version.Keep using the standard Qwen-Image 2.1 text encoder, VAE, conditioning, sampler, and workflow components around it.<h2 id="important">Important</h2>These are unofficial community quantizations.GGUF support for image diffusion models is still evolving. Make sure ComfyUI and ComfyUI-GGUF are current enough to support <code>qwen_image</code>.Lower-bit quantization can affect:<ul><li>fine detail</li><li>typography</li><li>prompt adherence</li><li>edit fidelity</li><li>anatomy</li><li>texture consistency</li></ul>If a lower quant produces obvious degradation, move up one quant level before changing the rest of your workflow.<h2 id="credits">Credits</h2><ul><li>Qwen — Qwen-Image / Qwen-Image 2.1</li><li>Comfy-Org — ComfyUI-compatible Qwen-Image 2.1 model packaging</li><li>City96 / ComfyUI-GGUF — GGUF loading and quantization support for ComfyUI</li><li>llama.cpp — GGUF quantization infrastructure</li></ul>Quantizations / conversion work by RealRebelAI.<h2 id="links">Links</h2><ul><li>Qwen-Image 2.1 model files: <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Qwen-Image-2.1">https://huggingface.co/Comfy-Org/Qwen-Image-2.1</a></li><li>ComfyUI-GGUF: <a target="_blank" rel="ugc" href="https://github.com/city96/ComfyUI-GGUF">https://github.com/city96/ComfyUI-GGUF</a></li></ul>

Files with SHA256: 58790d3fb7ad48d7249b5de68bcdbb03c0e59d3d007f76799890578495d3fe56