ComfyUI Node
Load OmniGen Model
A ComfyUI node in loaders with 5 inputs and 1 output.
Load OmniGen Model
- OMNIGEN_MODEL
◄model_name▾►
◄weight_dtype▾►
◄store_in_vramfalse►
◄separate_cfg_infertrue►
◄offload_modelfalse►
Categoryloaders
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| model_name | COMBO | 1 options: none | |
| weight_dtype | COMBO | 4 options: default, fp8_e4m3fn, fp8_e4m3fn_fast, fp8_e5m2 | |
| store_in_vram | BOOLEAN | false | Keep model in VRAM between generations. Faster but uses more VRAM. |
| separate_cfg_infer | BOOLEAN | true | Use separate inference process for different guidance. Reduces memory cost. |
| offload_model | BOOLEAN | false | Offload model to CPU. Reduces VRAM usage but slower. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| OMNIGEN_MODEL | OMNIGEN_MODEL | — |