mi·do·naミ・ド・ナ

Home / Library / Help / Set up a runner / LM Studio

Guide

Set up LM Studio

LM Studio runs a language model on your own machine and serves it over a local API that mi·do·na can talk to. Nothing leaves your computer. This is the short version, enough to get a model answering. For the deeper details, LM Studio's own docs are excellent.

Get a model running

  1. Download & install LM Studio

    Grab the build for your OS from lmstudio.ai and install it like any other app. macOS, Windows, and Linux are all supported.

  2. Download a model

    Open the Discover tab (the magnifying glass) and search for a model. If you're unsure, start with a small, well-known instruct model that fits your RAM. LM Studio flags which quantizations your machine can handle.

    Bigger models are smarter but heavier. When in doubt, pick something in the 7-8B range to start.

  3. Load the model

    Once it's downloaded, load it so it's held in memory and ready to answer. The model has to be loaded before the server can serve it.

  4. Start the local server

    Go to the Developer (server) tab and start the server. By default it listens on port 1234 and serves an OpenAI-compatible API at http://localhost:1234/v1.

  5. Enable CORS

    In the server settings, turn on CORS. This lets a web app like mi·do·na, running in your browser, make requests to the local server. Without it the browser blocks the connection.

Next: connect mi·do·na

With the server running and CORS on, LM Studio is ready. Open the model connection panel in mi·do·na, choose LM Studio, point it at http://localhost:1234/v1, and load models.

Served over HTTPS and can't reach localhost? The browser can block an HTTPS page from talking to http://localhost - most often in Safari. You'll need a trusted HTTPS proxy in between. See

Reaching a local model over HTTPS

.

This is a quick-start. For anything deeper, see the official LM Studio documentation at lmstudio.ai/docs.