↓ Skip to main content
Chandler Thompson
  1. Posts/

Level up your local AI: getting started with LM Studio

·606 words·3 mins·
AI Productivity LM Studio Ollama
Table of Contents

LM Studio is a desktop app for running local large language models. It handles model downloads, prompt chat, and provides a local OpenAI-compatible API endpoint other tools can hit. The main reason to use it: you keep your prompts on your own machine instead of routing them through someone else’s API.

This guide covers installing LM Studio, downloading models inside the app, linking in models already managed by Ollama via Gollama, and a quick tour of the interface.

Installing LM Studio with Homebrew on macOS
#

On macOS, install with Homebrew:

brew install lmstudio

Homebrew handles the cask download and installation. Launch from your application menu or Spotlight.

Installing LM Studio via direct download
#

If you prefer not to use Homebrew or other package managers, you can install LM Studio directly from its official website. Here’s how:

  1. Navigate to the LM Studio download page.
  2. Click on the appropriate installer for your OS (macOS, Windows, or Linux).
  3. Double-click the downloaded file and follow the installation wizard.
  4. Once installed, launch LM Studio through your application launcher.

Installing models directly in LM Studio
#

If you’d rather keep everything in one place, LM Studio allows you to download and manage models directly. Here’s how:

  1. Go to Settings -> Models in LM Studio.
  2. Click Add New Model or Browse to explore available options.
  3. Select and download the model you want to use.

It’s that straightforward. No extra tools or steps required.

Connecting local models via Ollama
#

If you’ve been using Ollama to manage your local language models, you can integrate them into LM Studio using the gollama tool. gollama is a macOS/Linux utility that provides a text-based user interface for managing Ollama models and can link them to LM Studio. Here’s how to set it up:

  1. Install gollama:

    go install github.com/sammcj/gollama@latest

  2. Run gollama:

    gollama

  3. Link Models to LM Studio:

    • Within the gollama interface, select the models you want to link to LM Studio.
    • Use the provided key bindings to link the selected models.

By following these steps, your Ollama-managed models will be accessible within LM Studio’s interface, allowing for seamless interaction.

Note: Ensure that both Ollama and LM Studio are properly installed on your system before proceeding with the above steps.

LM Studio basics
#

Now that you’re set up, let’s explore some core features of LM Studio:

1. Model selection
#

Switch between different models in the main interface. This makes it easy to compare performance or choose the best model for a specific task.

2. Prompt & chat
#

A chat-style interface for sending prompts and reading responses. Useful for testing a model before wiring it into something else.

3. Token usage & performance stats
#

Built-in token counts and per-response timing. Worth checking when comparing models or tuning context-window settings.

4. Advanced settings & fine-tuning
#

The settings menu exposes temperature, top-p, context length, and other generation parameters. Defaults are reasonable; you’ll want to adjust these once you have a sense of how a given model behaves.

Closing notes
#

LM Studio is the most straightforward way I’ve found to run local language models without giving up a chat-style interface or having to think about server processes. The Ollama integration via Gollama means you don’t have to pick between the two — both can coexist, and LM Studio reads from Ollama’s model store directly.

If you’re already running models via Ollama at the command line, LM Studio adds a UI on top without forcing you to migrate. If you’re starting fresh, it’s a reasonable first stop before deciding whether you need the CLI as well.

Additional resources
#

Chandler Thompson
Author
Chandler Thompson
Perpetual Hobbyist.

Related