ComfyUI Node
LlamaCPP Preload Model
Kick off a llama.cpp model load in the background so the first Chat request is fast. Optionally clears ComfyUI's VRAM first so the model has room to load. Runs on every execution and passes through immediately without waiting for the load to finish.
LlamaCPP Preload Model
- connectivity
- anything
- output
◄clear_vramtrue►
CategoryLlamaCPP API
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| connectivity | LLAMACPP_CONNECTIVITY | Wire a LlamaCPP Connectivity node here. | |
| anything | * | Passthrough. The load fires when this node runs, so wire it where you want the model warm. | |
| clear_vram | BOOLEAN | true | Unload all ComfyUI models and empty the GPU cache first (like easy-use's Clean VRAM Used), freeing VRAM for the LLM. Turn off when the model fits. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| output | * | — |