UNet Name (GGUF)

If you run Flux, SD3.5, or a big video model on a normal consumer card, you probably run it as GGUF - the quantized format that made 12B+ models fit in 12GB by trading a tiny bit of quality for a huge chunk of VRAM. The GGUF loader gives you a patched MODEL and its filename, but if you want that filename as text - for auto-named outputs, prompt templates, or just knowing which quantization you used - the loader doesn't hand it over. UNet Name (GGUF) (class UNETNameGGUF) is the lhyNodes utility for exactly that: a dropdown of every .gguf UNet in your models/unet_gguf folder, with the selected name as the only output.

It's the GGUF-flavored sibling of the pack's plain UNETName node. Same idea, different folder: the plain one reads diffusion_models, this one reads unet_gguf - the folder where GGUF-format UNets live. So if you're running a Q8 or Q4_K_M Flux UNet, this is the name node that sees it. The output is the filename as a string (typed *, so it plugs into any text socket), ready to feed a String Format node for filenames or a logging line.

The inputs that matter

• unet_name - the dropdown of GGUF UNet files in models/unet_gguf. That's the whole thing.

Output: unet_name, the selected file as a string.

Where it earns its keep

For GGUF users, knowing which quant you generated with becomes part of your record-keeping. Q8 is basically fp16 at half the size, Q5 is the last stop before visible loss, Q4_K_M is the accepted compromise - and if you're A/B testing quantization levels, having the exact filename ride along into your output naming is genuinely useful. Wire UNETNameGGUF into a String Format template, and your saved files start telling you what they were made from.

Two practical notes. First, if your GGUF isn't in the dropdown, check the folder - it has to live in models/unet_gguf (that's the standard GGUF pack convention), not in diffusion_models. Second, don't confuse this with the plain UNETName: a regular safetensors UNet won't appear here, and a GGUF won't appear there. Use the node that matches what you're loading.

Installing it

Part of lhaoyun6/ComfyUI-lhyNodes. Install via ComfyUI Manager by searching for lhyNodes, or:

cd ComfyUI/custom_nodes
git clone https://github.com/lihaoyun6/ComfyUI-lhyNodes.git
python -m pip install -r ComfyUI-lhyNodes/requirements.txt

Restart ComfyUI. The pack's shared requirements are ultralytics, opencv-python, numpy, and yarl; this node needs no models of its own - it just lists the ones you already have.

It's a two-minute addition that pays off the first time you stare at an output folder and can't remember which quant made which image. If GGUF is how you run things, this is the name node you want.