Models Registry
Quantized GGUF builds of diffusion and audio models, maintained by molbal.
| Model | Focus | Repository |
|---|---|---|
| Fibo 2.0 | Text-to-image and image editing | HF |
| Flux 3 Klein | Text-to-image and multi-reference editing | HF |
| MiniMax Music 3 | Audio and music up to 5 minutes | HF |
| MiniMax H3 Turbo | Video and multimodal | HF |
| LTX 2.5 | Open-weights video generation | HF |
| MiniMax H3 | Omni-modal generation | HF |
| Krea 2 | Image synthesis and design | HF |
| Ideogram 4 | Text rendering and graphic design | HF |
| Qwen Image 2.1 | Text-to-image and image editing | HF |
| Qwen Image 2.1 Turbo | Turbo text-to-image and image editing | HF |
| MOSS-Image | Text-to-image with SFT and 8-step Turbo quants | HF |
| Kromoss | MOSS-Image-Base fine-tune for text-to-image | HF |
How to use these models
- Install ComfyUI with the comfyui-gguf-reboot node pack.
- Download the GGUF files from the model's Hugging Face repository.
- Pick the quant format that matches your VRAM budget. See Quant Formats.
- Load the model with the GGUF loader node and use a workflow from Workflows.
Not sure which quant to choose?
Read the per-model pages or Quant Formats.
Requesting a new model
File a request at github.com/molbal/gguf. See Community for details.