Local Ollama Model Selector

Local Ollama Model Selector is the smallest node in the Local Ollama Translator pack, and it does exactly one useful thing: it turns the list of models Ollama actually has on your machine into a dropdown, then hands you the one you picked as a plain string. No more typing qwen3:8b by hand into a text box and praying you didn't fat-finger it.

It lives in the Local/Ollama menu, right next to the translator it exists to feed. The node it pairs with - Local Ollama Translator Multilingual - takes its model name as a plain STRING widget, so you can just type the name. This node exists for two cases: you juggle a few models (an 8B for quick translations, a 14B when a passage fights back) and want a picker instead of a memory exercise, or you're building a workflow you might run on a machine with a different set of installed models and you want it to adapt instead of hard-failing.

How it works

On load, the node fires a GET at http://127.0.0.1:11434/api/tags - the same endpoint ollama list hits - and fills the dropdown from the JSON. That's the whole mechanism, and it's where the one real gotcha lives: if Ollama isn't running (or the request times out after 1.5 seconds), the code silently falls back to a hardcoded list of qwen3:8b, qwen3:14b, qwen2.5:7b, and pure-zh:latest. So the dropdown showing a model you've never pulled is your tell that Ollama was down when ComfyUI loaded. It degrades gracefully, but it degrades with a lie.

Inputs and outputs

One required input, one output. That's it.

model - an enum of the discovered model names. Pick the one you want.
model (output, STRING) - the name you selected. Wire it into the Multilingual translator's model_override input, or into anything else that eats a model-name string.

It's not an output node - nothing renders, it just passes a string along the graph.

Installing and using it

Same as the rest of the pack. Easiest route is ComfyUI Manager: search for Local Ollama Translator, install, restart ComfyUI. Or clone it straight into custom_nodes:

cd ComfyUI/custom_nodes
git clone https://github.com/InfernusIntraMe/ComfyUI-Local-Ollama-Translator.git

Then restart ComfyUI and hard-refresh the browser with Ctrl+F5. The pack has no Python dependencies - it uses only the standard library, which is refreshingly boring in a category full of packs that drag in half of PyPI. What it does need is Ollama itself running, with a model already pulled:

ollama pull qwen3:8b
ollama list

The gotchas

The dropdown only refreshes when the node loads, which happens at ComfyUI startup. Pull a new model mid-session and it won't appear until you restart ComfyUI - the README is explicit about this, and it's the #1 "it's broken" report for this node. It isn't broken, it just doesn't poll.

Honest take: this node is a nicety, not a requirement. The translator works fine with the model name typed into its widget, and the pack even defaults to qwen3:8b so you may never touch this node at all. Reach for it when you're juggling multiple local models or sharing workflows between machines. If you only ever translate with one model, skip it and spend the canvas space on something that earns it.