Flux GGUF Quantized Weights in ComfyUI
Quantized transformer weights that let 8 GB and 12 GB Windows GPUs run Flux.1 Dev and Hunyuan Video without swapping to disk. Requires 6 GB–12 GB depending on Q4_K_S to Q8_0 VRAM and loads from models/unet/.
Updated Oct 17, 2026 · 1 min read
- Model Family
- City96 ComfyUI-GGUF
- VRAM Target
- 6 GB–12 GB depending on Q4_K_S to Q8_0
- ComfyUI Folder
- models/unet/
- Precision
- GGUF Q4_K_S, Q5_K_M, Q8_0
- Default Sampler
- UnetLoaderGGUF + DualCLIPLoaderGGUF
Where to place Flux GGUF Quantized Weights weights on Windows
Quantized transformer weights that let 8 GB and 12 GB Windows GPUs run Flux.1 Dev and Hunyuan Video without swapping to disk.
Place the primary `.safetensors` or `.gguf` weight file inside `models/unet/` inside your ComfyUI directory, or point extra_model_paths.yaml to your existing shared models drive.
After moving the file, press **R** on the ComfyUI Studio canvas (or click Refresh) so the loader dropdown lists the new checkpoint immediately without restarting the server.
Recommended loader nodes and sampler settings
- **Precision & Memory:** GGUF Q4_K_S, Q5_K_M, Q8_0. Cards at the lower end of 6 GB–12 GB depending on Q4_K_S to Q8_0 should pair this model with FP8 or GGUF quantization and VAE Decode (Tiled).
- **Sampler & Scheduler:** UnetLoaderGGUF + DualCLIPLoaderGGUF. Wire the model output into KSampler and connect the matching CLIP and VAE streams.
- **Installer:** If you have not set up the desktop environment yet, get the Windows build at /download.
Flux GGUF Quantized Weights Quick Reference for ComfyUI Windows
| Parameter | Recommended Setting | Notes |
|---|---|---|
| Target Directory | models/unet/ | Refresh canvas with R key after copying |
| VRAM Tier | 6 GB–12 GB depending on Q4_K_S to Q8_0 | Use --lowvram or GGUF Q4 on 8 GB cards |
| Weight Format | GGUF Q4_K_S, Q5_K_M, Q8_0 | Never load untrusted .ckpt pickle files |
| Sampler Recipe | UnetLoaderGGUF + DualCLIPLoaderGGUF | Wire through core KSampler node |
FAQs
Which folder does Flux GGUF Quantized Weights go into in ComfyUI?
Put Flux GGUF Quantized Weights inside `models/unet/` in your ComfyUI installation, then press R on the canvas to refresh node dropdowns.
Can I run Flux GGUF Quantized Weights on an 8 GB NVIDIA GPU?
Yes — use FP8 or GGUF Q4/Q5 weights via ComfyUI-GGUF and keep tiled VAE decoding enabled so peak VRAM stays under 8 GB.
Run on your hardware. Free forever.
The full ComfyUI engine: open source, fast, extensible, and yours to run however you want. Download the desktop app and open this workflow on your own machine.
