How to Set Up LM Studio for Local AI

Download LM Studio, grab a model from the built-in hub, start the local server, and have a working private AI chat running on your own hardware in about ten minutes.

10 minBeginner

LM Studio is a free desktop app that lets you download and run large language models entirely on your own machine. Nothing leaves your computer, there is no API key to manage, and it works offline. This guide takes you from a fresh download to a working chat session.

What you need

A Mac (Apple Silicon or Intel), Windows PC, or Linux machine
At least 8 GB of RAM (16 GB recommended for 7B models)
4 GB of free disk space for a small model, 8 GB for a 7B model
LM Studio downloaded from lmstudio.ai

Step 1: Download and install LM Studio

Go to lmstudio.ai and click the download button for your operating system. On macOS, drag the app to Applications. On Windows, run the installer. On Linux, download the AppImage, make it executable, and run it.

Linux — make AppImage executable

$chmod +x LM_Studio-*.AppImage

$./LM_Studio-*.AppImage

Step 2: Browse and download a model

Open LM Studio and click the Discover tab (the magnifying glass icon on the left). This connects to Hugging Face and shows popular models filtered by your hardware. For a first run, search for phi-3.5-mini and download the Q4_K_M variant — it weighs about 2.2 GB and runs on any modern laptop.

LM Studio — Discover tab

[Discover] [Chat] [Local Server] [My Models]

-----------------------------------------------

Search: phi-3.5-mini

microsoft/Phi-3.5-mini-instruct

Q4_K_M 2.2 GB [Download]

Q8_0 4.1 GB [Download]

mistralai/Mistral-7B-Instruct-v0.3

Q4_K_M 4.1 GB [Download]

Search results showing download size and quantization variant.

Pick the right quantization

Q4_K_M is the best starting point for most people — good quality at half the file size. Use Q8 if you have spare VRAM and want higher accuracy. Avoid Q2 and Q3 unless you are severely constrained.

Step 3: Chat with the model

Click the Chat tab (the speech bubble icon). Select your downloaded model from the model picker at the top, then type a message. The first response takes a few seconds to load the model into memory; subsequent replies are much faster.

LM Studio — Chat

You

Explain what a transformer model is in two sentences.

Agent

A transformer model is a neural network architecture that processes sequences by attending to all positions simultaneously rather than reading left-to-right. This attention mechanism lets it capture long-range dependencies and is the foundation behind almost every modern LLM.

Step 4: Start the local server

Click the Local Server tab and press Start Server. LM Studio exposes an OpenAI-compatible API at http://localhost:1234. Any app that works with the OpenAI SDK (Python, Node, curl) can point at this URL instead, with no API key required.

Test the local server with curl

$curl http://localhost:1234/v1/chat/completions \

-H "Content-Type: application/json" \

-d '{

"model": "local-model",

"messages": [{"role": "user", "content": "Hello"}]

Returns a standard OpenAI-format JSON response

Context length setting

LM Studio defaults to a 4096-token context window. For longer documents, open Model Settings in the chat sidebar and increase Context Length — though each doubling roughly doubles the RAM usage.

Result

You now have a local AI chat running on your own machine, a server that any OpenAI-compatible client can call at localhost:1234, and a model library you can browse and swap without a subscription.

Local models run better with more VRAM. CompareRTX GPUs on Amazonbefore you upgrade.(affiliate link. We may earn a commission at no extra cost. Disclosure)