Load Remote Models🧸

If you connect Advanced Prompt Enhancer to a service like OpenRouter, you have to tell it which models exist, and OpenRouter has hundreds. Typing model names into a text file by hand is miserable. Load Remote Models does it for you: it queries the service's model list, filters it, and dumps the results straight into the dropdown APE reads from. One click instead of copy-pasting model IDs off a webpage.

Niche node, but a real quality-of-life win if you live in the world of aggregator LLM services.

How it works

You pick a service, give it include/exclude filters, and it fetches that service's current model catalog over the API and writes the matching names into Plush's opt_models.txt file. That file is what populates the Optional_model dropdown on Advanced Prompt Enhancer, so after you run this node, your filtered models are just there, ready to select. Right now it supports OpenRouter, APIpie, NanoGPT, and ArlAI; the author has been adding services over time.

The inputs that matter

• Service_Name: which aggregator to pull from (OpenRouter, APIpie, NanoGPT, ArlAI).
• Include_Filter / Exclude_Filter (multiline): text filters to narrow the list. This is the important part, because you do not want all of OpenRouter's models in your dropdown. Filter to, say, the vendor or size you care about.
• Output_to: either Test_Output (just show you the list so you can preview what the filters catch) or Optional Models List (actually write it into the file APE uses). Run it on Test_Output first to sanity-check your filters, then switch to write.
• Remove_Prior_Service_Name_Entries (boolean, default true): clear out the previous entries for this service before writing, so the list doesn't accumulate stale names.
• Sort_Models (boolean, default true): alphabetize the result.
• Custom_ApiKey (optional): most of these services need a key, supplied through the Custom API Key node.

Outputs are Test_Output (the model list as text) and Troubleshooting. This is an output node, so it runs and writes as an endpoint.

Installing it

Ships with Plush-for-ComfyUI. Install via ComfyUI Manager (search "Plush", install Plush-for-ComfyUI, restart), or manually: cd ComfyUI/custom_nodes && git clone https://github.com/glibsonoran/Plush-for-ComfyUI.git, then pip install -r requirements.txt, restart.

Common issues

You'll need an API key for the service, read in via Custom API Key (with your key stored in an environment variable). If the fetch returns nothing, check the key first, then loosen your filters, since an over-tight include filter can match zero models. Preview with Output_to set to Test_Output before you write, so you're not surprised by what lands in the dropdown. And remember the models only show up in Advanced Prompt Enhancer's Optional_model field, and only apply when you're using one of the (URL) AI services, so if you selected a model but APE ignores it, make sure you're on a URL-based connection. The Troubleshooting output reports the real error when the API call fails.