How to install and set up LM Studio on Mac, Windows and Linux

LM Studio is a free application that lets you use artificial intelligence models (like the ones behind ChatGPT) directly on your computer, without sending your text to any outside server and without paying for a subscription. In this guide you will install it from scratch, download your first model, chat with it and, at the end, get it ready so that other applications can use it too.

You don't need to know how to program. Every step explains what to click and what you should see on screen.

What LM Studio is and what you will achieve

A language model (or LLM, for "large language model") is a large file containing an artificial intelligence that can understand and write text. Companies such as Google, Alibaba and OpenAI publish some of their models so that anyone can download them.

LM Studio does three things for you:

• It finds and downloads those models with one click.
• It gives you a chat window, similar to ChatGPT, to talk to them.
• It can run as a local server: a service that other applications on your computer (for example Open WebUI or LiteLLM, covered in other guides) can use.

By the end of this guide you will have an AI assistant that works even without an Internet connection.

Before you start: check your computer

AI models need memory. What matters most is your computer's RAM.

Mac

• You need a Mac with an Apple chip (M1, M2, M3, M4 or later). Macs with an Intel processor are not supported.
• You need macOS 14 Sonoma or later.
• 16 GB of memory or more is recommended. It works with 8 GB, but only with small models.

To check, click the Apple menu (the apple in the top-left corner) and choose About This Mac. You will see the chip, the memory and the macOS version.

Windows

• Windows 10 or 11, with a modern processor (64-bit Intel or AMD, or a laptop with a Snapdragon chip).
• 16 GB of RAM is recommended. A graphics card with 4 GB or more of its own memory helps a lot, but is not required.

To check, press the Windows + Pause keys together, or open Settings → System → About. Look for the "Installed RAM" line.

Linux

• A recent 64-bit distribution. LM Studio is mainly tested on Ubuntu 20.04 and 22.04; newer versions usually work but are less tested.
• 16 GB of RAM is recommended.

Disk space

Each model takes between 2 and 20 GB. Keep at least 20 GB free to start comfortably.

Step 1. Download LM Studio

1. Open your browser and go to lmstudio.ai/download.
2. The page detects your system and offers the right download. If it doesn't, choose your system from the list.
3. Click the download button and wait for it to finish. The file is a few hundred megabytes.

Step 2. Install it

On Mac

1. Open your Downloads folder and double-click the file ending in .dmg (for example, LM-Studio-0.4.25-1-arm64.dmg).
2. A window opens with the LM Studio icon and a shortcut to the Applications folder. Drag the LM Studio icon onto the Applications folder.
3. Open the Applications folder and double-click LM Studio.
4. If macOS asks whether you want to open an application downloaded from the Internet, click Open.

On Windows

1. Open your Downloads folder and double-click the file ending in .exe (for example, LM-Studio-0.4.25-1-x64.exe).
2. If a blue Windows protected your PC window appears, click More info and then Run anyway. This happens with many newly released programs.
3. Follow the installer with the options already selected.
4. When it finishes, open LM Studio from the Start menu.

On Linux

LM Studio comes as an AppImage: a single file that runs without installing anything.

1. Open the Terminal (on Ubuntu, press Ctrl + Alt + T).
2. Go to your downloads folder by typing this command and pressing Enter:

cd ~/Downloads

3. Give the file permission to run (the asterisk completes the name for you):

chmod +x LM-Studio-*.AppImage

4. Open it:

./LM-Studio-*.AppImage

Step 3. First launch

The first time, LM Studio shows a few welcome screens. You can read them or skip them.

Look for the mode selector, usually at the bottom of the window: it has two options, User and Developer. To start, leave it on User: this mode is designed for beginners and hides the advanced options. We will switch to Developer later, when we turn on the server.

Step 4. Download your first model

1. In the left sidebar, click the magnifying glass icon (it's called Discover). You can also press ⌘ + 2 on Mac or Ctrl + 2 on Windows and Linux.
2. In the search box, type the name of the model you want. These are good options to start with, depending on your computer's memory:

• 8 GB of memory: gemma-3-4b (about 2.4 GB) or qwen3.5-2b (about 2 GB).
• 16 GB of memory: gemma-4-e4b (about 5.9 GB) or qwen3.5-9b (about 7 GB).
• 32 GB or more: gpt-oss-20b (about 12 GB) or qwen3.8-27b (about 16 GB).

3. Click the model you chose. Its variants appear on the right. LM Studio recommends downloading a 4-bit or higher version: look for names like Q4_K_M. If the app marks an option as recommended for your computer, choose that one.
4. Click Download and wait. You can follow the progress at the bottom of the window.

Step 5. Load the model and chat

1. Click the Chat icon in the sidebar.
2. At the top of the window you will see the model selector ("Select a model to load"). Click it and choose the model you just downloaded.
3. Wait a few seconds while it loads into memory. When it's done, you will see its name at the top.
4. Type a message in the box at the bottom, for example "Explain in three sentences what a language model is", and press Enter.

To start a new conversation, press ⌘ + N (Mac) or Ctrl + N (Windows and Linux).

Congratulations: you now have an AI running on your computer. From here on you don't need the Internet to use it.

Step 6. Useful settings to start with

You don't need to change anything to use LM Studio, but it's worth knowing three settings. They appear when you load a model (open the advanced options in the selector), or as default values for each model in the My Models section of the sidebar, by clicking the gear next to the model.

• Context Length: how much text the model can "remember" in a conversation. A higher value allows longer conversations and documents, but uses more memory. If the app is slow or reports memory errors, lower it.
• GPU Offload: how much of the model runs on the graphics card, which is much faster. If you leave it alone, LM Studio decides automatically.
• Models folder: by default models are saved in the .lmstudio/models folder inside your user folder. If you are short on space, you can choose another location in My Models, such as an external drive.

Step 7. Turn on the local server

This step is optional, but it is what lets other applications use your models: Open WebUI, LiteLLM, code editors or agents such as Hermes. You will need it in the next guides of this series.

1. Turn on the advanced mode: in the mode selector, switch from User to Developer. You can also do it in Settings → Developer.
2. A Developer tab appears in the sidebar. Open it.
3. Turn on the Start server switch. Once it's running, you will see the server address: http://localhost:1234.

Check that it works

Open your browser and type this address:

http://localhost:1234/v1/models

You should see text with braces and quotes (JSON format) that includes the names of your models. That means the server is answering.

The address other applications will use is:

http://localhost:1234/v1

Write it down: you will need it in the LiteLLM, Open WebUI and Hermes guides.

Using it from other devices on your network

In the Developer tab, under Server Settings, there is an option called Serve on Local Network. If you turn it on, other computers in your home or office can use the server with your computer's IP address instead of localhost.

Extra: the lms command-line tool

LM Studio includes a terminal program called lms. You don't need it to use the app, but it lets you do the same things by typing commands, which is handy for automation.

To try it, open LM Studio at least once and then open a terminal:

• Mac: press ⌘ + Space, type "Terminal" and press Enter.
• Windows: open the Start menu, type "PowerShell" and press Enter.
• Linux: press Ctrl + Alt + T.

Type:

lms --help

If you see a list of commands, you're set. If you get a "command not found" error, run this and open a new terminal:

npx lmstudio install-cli

That last command requires Node.js. If you don't have it, don't worry: the lms tool is optional.

The most useful commands are:

• lms ls: lists the models you have downloaded.
• lms get google/gemma-3-4b: downloads a model.
• lms load <model>: loads a model into memory. With --context-length 16000 you set its context length.
• lms ps: shows which models are loaded.
• lms server start: starts the local server without opening the window.
• lms server stop: stops it.

Common problems

• On Mac, LM Studio won't open or says it isn't compatible. Check that your Mac has an Apple chip (M1 or later) and macOS 14 or later.
• The model download gets stuck. Check your connection and free disk space. You can pause and resume it from the bottom of the window.
• The model is very slow or the computer freezes. The model is too big for your memory. Download a smaller one or lower the Context Length.
• The browser can't open localhost:1234. Check in the Developer tab that the Start server switch is on and that the port is 1234.
• Another application says it can't find the model. With the server running, load the model in LM Studio, or turn on Just in Time Model Loading in Server Settings so it loads automatically when another application asks for it.

Next steps

LM Studio is now up and running. The next guides in this series show how to install Ollama, another popular way to run models locally; how to bring all your models together in one place with LiteLLM; and how to use them from a full web interface with Open WebUI.

Next guide: How to install and set up Ollama on Mac, Windows and Linux