LM Studio

LM Studio can serve loaded models over an OpenAI-compatible API from its local server, on port 1234 by default.

Start the local server

In LM Studio, open the Developer tab, load a model, and start the server. The server page shows the base URL, usually http://127.0.0.1:1234.

If you turned on API-key protection in LM Studio, keep that key at hand; otherwise the server accepts requests without one.

Add the upstream

noket upstream add \
  --name lmstudio \
  --protocol openai \
  --url http://127.0.0.1:1234 \
  --model <model-id>

The model id is the identifier LM Studio shows in the server’s model list, which is usually the name of the loaded model. The quickest way to get it right is the interactive form, which reads the list from the server and lets you pick:

noket upstream add

Then verify and start:

noket upstream test lmstudio
noket up

Keep the server running

LM Studio stops its server when you quit the app, so the upstream works only while LM Studio is open. If you want your models available whenever the machine is on, use a server that runs as a service — vLLM or Ollama — or keep LM Studio running in the background.

Next

• Create a model in the console
• vLLM