Unet Loader (GGUF, Dynamic VRAM):
UnetLoaderGGUFDynamicVRAM loads a GGUF diffusion model using the dynamic VRAM loading path, preparing it for use in a ComfyUI workflow. GGUF models can contain quantized weights, which store model data in a more compact form; the loader uses GGUF-aware operations so the model can work with those weights. The dynamic path is intended to manage model memory as it runs, which may help when GPU memory is limited, but it does not guarantee that every model will fit. The node also displays loading progress while it reads the model file and reports an error if ComfyUI cannot identify the model type.
Unet Loader (GGUF, Dynamic VRAM) Input Parameters:
unet_name
Selects the GGUF diffusion model file to load. The available choices come from the installed GGUF model file list, so the exact options depend on the files available in your ComfyUI setup. Your choice determines which model the node reads and therefore which model your downstream workflow uses. This is a required selection with no default, minimum, or maximum specified in the provided source.
Unet Loader (GGUF, Dynamic VRAM) Output Parameters:
model
A ComfyUI MODEL object representing the loaded diffusion model. Connect it to a downstream node that accepts a MODEL input, such as the model input of a sampling workflow. This output is the prepared model, not a finished image. The loader returns one model object when loading succeeds; if loading fails, it raises an error instead of producing a usable output.
Unet Loader (GGUF, Dynamic VRAM) Usage Tips:
- Choose the intended GGUF model from
unet_namebefore running the workflow. If the model you expect is absent from the menu, check that ComfyUI can discover its file in the configured GGUF model locations. - Connect
modelto a node that accepts aMODELinput. You will still need the other components required by your image-generation workflow; this node loads the diffusion model only. - Watch the loading progress indicator when opening a large model. The node tracks progress through the selected file, so allow loading to complete before diagnosing a stalled downstream workflow.
- If you are troubleshooting memory use, try a model that requires less memory. Dynamic VRAM loading is enabled by this node and has no exposed setting to adjust or disable here.
Unet Loader (GGUF, Dynamic VRAM) Common Errors and Solutions:
ERROR: Could not detect model type of: <file path>
- Explanation: This message is raised when ComfyUI cannot identify the loaded file as a supported diffusion model.
- Solution: Check that you selected a GGUF diffusion model intended for this loader, rather than another kind of GGUF file. Confirm that the file downloaded completely and that your ComfyUI installation supports the model architecture.
Model file not found or could not be opened
- Explanation: The provided source does not specify an exact message for this case, but loading can fail if the selected file has been moved, deleted, or made inaccessible since it appeared in the menu.
- Solution: Confirm that the file still exists in a configured model location, is readable, and is selected under its current name. Refresh the model list or restart ComfyUI after moving files.
Dynamic VRAM requirement error
- Explanation: Before loading, this node checks that its dynamic VRAM requirements are available. The provided source does not show the check’s exact error message or requirements.
- Solution: Read the full error in the ComfyUI console, then check the installation and compatibility instructions for the node package and your ComfyUI version. Do not assume that changing the model file alone will resolve a failed prerequisite check.
GPU out-of-memory error
- Explanation: Dynamic VRAM loading does not guarantee that a model or the rest of a workflow will fit in available GPU memory. The exact error wording depends on where the failure occurs.
- Solution: Try a smaller model or reduce memory demands elsewhere in the workflow, such as image dimensions or batch size. Close other applications using GPU memory, then retry.
GGUF file loading or decoding error
- Explanation: The GGUF loader can fail while reading a damaged, incomplete, or incompatible file. The provided source does not specify one fixed error message for these failures.
- Solution: Verify that the selected file is a complete GGUF diffusion model download. If it appears damaged, obtain a fresh copy, and use the console traceback to identify any more specific compatibility issue.
