How to Use LM Studio: Run AI Models on Your Own Computer

LM Studio is the friendliest way I know to run an AI chatbot on your own computer. There's no terminal and no code: you install an app, pick a model from a built-in browser, and start chatting. For anyone curious about private, offline AI, it's a lovely place to start.

This guide covers what LM Studio is, which computers it runs on, how to download your first model, how to chat with your own documents, and the more advanced features worth knowing. Every detail comes from LM Studio's own documentation.

What LM Studio is

LM Studio is a desktop app for running large language models, the type of AI behind chatbots, on your own machine. Its docs mention models such as Llama, DeepSeek and Qwen. Instead of sending your questions to a company's servers, the model runs on your laptop or desktop.

The big difference from command-line tools is the interface. LM Studio looks and feels like a normal chat app, with a model search screen, a chat window and settings you can click through.

Which computers can run LM Studio

LM Studio's documentation says it's available for macOS, Windows and Linux. It generally supports:

• Apple Silicon Macs.
• Windows PCs with x64 or ARM64 processors.
• Linux PCs with x64 processors.

Under the hood, it uses a popular engine called llama.cpp for models in the GGUF format on all three systems, and Apple's MLX framework on Apple Silicon Macs. You don't need to understand either; LM Studio picks what fits.

The size of model you can run comfortably depends on your computer's memory. Check LM Studio's system requirements page and start with a smaller model if you're unsure.

How to install LM Studio and download your first model

1. Go to the downloads page on LM Studio's website and install the version for your computer.
2. Open LM Studio and go to the model search area.
3. Search for a model. LM Studio searches and downloads models through Hugging Face.
4. Download a model that suits your computer. The docs point out you need to download the model files before you can chat.
5. Load the model and start typing in the chat window.

Once you have a model downloaded, LM Studio says it can run entirely offline, which is handy on trains, planes or anywhere with patchy internet.

Chatting with your own documents in LM Studio

One of my favourite features is document chat. LM Studio's docs say you can attach documents to your chat messages and work with them entirely offline. This technique is often called RAG, short for retrieval-augmented generation, which simply means the AI looks things up in your files before answering.

• Ask questions about a long contract draft without uploading it anywhere.
• Summarise meeting notes or research papers.
• Pull key points from a policy document for your team.
• Check a proposal against your own guidelines.

As always, double-check anything important against the original document. Local models can make mistakes just like online ones.

LM Studio's local server and other advanced features

When you're ready to go further, LM Studio has extras that developers love:

LM Studio feature	What it lets you do
Local server	Serve models on OpenAI-like endpoints, on your computer or across your network
MCP support	Connect MCP servers so local models can use tools and data
Document chat	Ask questions about files offline
Headless mode (llmster)	Run LM Studio without the app window on servers
Model browser	Find and download models from Hugging Face

The local server means other apps that already work with OpenAI-style connections can often talk to LM Studio instead. If you're using AI to help with coding, my tips on speeding up coding work with AI prompts work with local models too.

Getting better answers from models in LM Studio

• Give the model a clear role and task, just as you would with an online chatbot.
• Keep prompts focused: local models tend to do better with one job at a time.
• Try two or three different models on the same question and keep the one that suits you.
• Start a fresh chat when you switch topics so old context doesn't confuse the answer.

LM Studio privacy and safety tips

Running a model locally is a strong privacy choice, because your chats stay on your machine. It's worth comparing with the settings you'd check on a hosted chatbot; my guide to whether ChatGPT is private covers that side.

• Download models from sources you trust, and check each model's licence before commercial use.
• If you turn on the local server across your network, think about who else can reach it.
• Only connect MCP servers you trust, and review what they can access.
• Review any AI-written code before you run it.

LM Studio versus online chatbots

Local models are getting better all the time, but the biggest online assistants are still more capable for many tasks. I'd use LM Studio for private drafting, document questions and experiments, and keep an online assistant for heavy research or complex writing. My comparison of Claude and ChatGPT helps if you're choosing one of those.

If you'd like help deciding whether local AI makes sense for your business, and setting it up sensibly, that's part of what I do as an AI consultant.

Frequently asked questions

What is LM Studio?

LM Studio is a desktop app for running large language models on your own computer, with a chat window and a built-in model browser.

Which models can LM Studio run?

LM Studio's docs mention models such as Llama, DeepSeek and Qwen, downloaded through Hugging Face. Check each model's licence before commercial use.

Does LM Studio work offline?

Yes. LM Studio says it can run entirely offline once you've downloaded model files.

Can LM Studio read my documents?

Yes. You can attach documents to chat messages and work with them offline, a technique often called RAG.

What is the difference between LM Studio and Ollama?

LM Studio is a desktop app with a visual model browser and chat window, while Ollama is mainly used from the command line and through its local API.