ComfyUI Node
Load SPEED Model
A ComfyUI node in video/SPEED with 2 inputs and 1 output.
Load SPEED Model
- model
◄model_pathspeed.pt►
◄precisionauto►
Categoryvideo/SPEED
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| model_path | COMBO | speed.pt | Checkpoint in models/speed-vfi/. The official ~447 MB checkpoint and pinned runtime source download on first use. |
| precision | COMBO | auto | Inference precision. Auto uses BF16 on supported CUDA GPUs, otherwise FP16. FP32 is slower and uses more VRAM. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| model | SPEED_VFI_MODEL | — |