ComfyUI Node
PromptForge Model
PromptForge model and runtime settings. One picker for clip / safetensors / GGUF; everything else is configured for the current GPU automatically.
PromptForge Model
- clip
- MODEL
- REPORT
◄model[safetensors] [DOWNLOAD] Qwen3-8B abliterated (HF) [8B]►
◄quantizationauto►
◄context_size0►
◄local_model_path►
◄mmproj_path►
Category🔮 PromptForge
Inputs (6)
| Name | Type | Default | Description |
|---|---|---|---|
| model | COMBO | [safetensors] [DOWNLOAD] Qwen3-8B abliterated (HF) [8B] | What to run. The list is split into local models (on disk or in the HF cache, ready to use; [LOCAL Q4/Q8] names the GGUF quants present) and [DOWNLOAD] catalog models that fetch on first use. |
| quantization | COMBO | auto | Quantization. auto picks the best local option for this GPU. |
| context_size | INT | 00–262144 | GGUF context length. 0 = auto (largest that fits, 16k first). |
| local_model_path | STRING | Optional local safetensors dir/file or GGUF file; overrides the picker. | |
| mmproj_path | STRING | Optional mmproj (.gguf) for GGUF image/video mode. | |
| clipopt | CLIP | Connected ComfyUI text encoder. Required when model is 'connected clip'. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| MODEL | PROMPTFORGE_MODEL | — |
| REPORT | STRING | — |