📋 Ollama Cloud Model List

This node doesn't translate a thing. Don't come looking for it to do anything clever - OllamaCloudModelList is the little debug sibling inside the Comfy_olima pack, and its whole job is to ask an Ollama server "what models can you run?" and dump the answer as plain text. It exists because the pack's real star, the Ollama Cloud Translator, ships a hardcoded model dropdown that goes stale. When you want to know what's actually available, you ask this node instead.

Why you'd reach for it

Say you're running the Translator in local mode against your own Ollama install. You've pulled qwen2.5:7b (or an abliterated Llama, the usual pick for local prompt work - see the LLM-in-ComfyUI pattern) and you want to feed it into the Translator's custom_model field. But you can't remember the exact tag, and typos get you a useless error string. Drop this node on the canvas, run it once, and the answer is sitting there in model_list.

It's equally useful to sanity-check that your Ollama Cloud API key is valid before you build a whole workflow on top of it. If the server says "no" to a GET /models with your key, it's going to say "no" to every translation call too - better to learn that from a one-second debug node than from a broken batch run.

How it works

Under the hood it's a single requests.get(f"{base_url}/models") call - the OpenAI-compatible models endpoint, the same one every Ollama server (and most LLM API gateways) exposes. Two inputs, both strings:

• base_url (required): defaults to https://ollama.com/v1 for cloud mode.
• api_key (optional): set it for cloud, leave blank for a local server like http://localhost:11434/v1.

It returns one string output, model_list: every model ID the server reports, joined with newlines. It's marked as an output node, so it also prints the list to your ComfyUI console as it runs. A failed call doesn't raise - you get an [Error] ... string back instead, which is annoying for debugging but does keep your graph from dying.

The trick that matters

The Translator node does not read this node's output. The dropdown on the Translator is a fixed list baked into the pack's source, not something this node populates. So the real workflow is: run this to see what exists, then paste the model name you want into the Translator's custom_model field. It's a two-step dance, but it's the only way to reliably use a model that isn't in the pack's hardcoded menu - and honestly, for a debug utility that's fine.

Install & troubleshooting

Install the whole pack, this node comes along free. No model downloads, no heavy deps - the pack's only requirement is requests, which ComfyUI already ships with.

• Empty or (empty) result: your server returned a 200 but zero models. Usually means you're hitting the wrong endpoint, or an auth-less server that wants a key.
• [Error] 401 on cloud: your api_key is wrong or expired. Recreate it at ollama.com/settings/keys.
• Connection refused on local: Ollama isn't running, or isn't exposing its API. ollama serve if you need it.

It's a 15-second sanity check that saves you a 15-minute debugging session. Not glamorous, but worth knowing where it lives.