LM Studio: run AI on your computer, offline (2026 guide)

What LM Studio is, real requirements on Mac, Windows and Linux, how to download a model and chat with your documents offline.

By Cognenza editorial team · Updated 16 October 2026 · 2 min read · Sources cited

LM Studio is a desktop app to download and run AI models on your own computer — Llama, Qwen, Mistral, gpt-oss and more. Your chats stay on your device. It’s the point-and-click alternative to Ollama.

Requirements (official docs)

• Mac: Apple Silicon (M1–M4), macOS 14+. 16 GB RAM recommended; 8 GB for small models. Intel Macs not supported.
• Windows: x64 with AVX2, or ARM (Snapdragon X Elite). 16 GB RAM and 4 GB+ dedicated VRAM recommended.
• Linux: x64 or ARM64 AppImage, Ubuntu 20.04+.

First steps

Download from lmstudio.ai, download a model in the app (via Hugging Face) and open a chat.

Chat with documents offline

Attach documents and work with them entirely offline (RAG) — handy for files you don’t want in the cloud.

Honest limits

Small local models are usually weaker than big cloud assistants and can hallucinate. Always verify.

In the glossary: Parameters.

In the glossary: Inference.

Advanced level: all the guides for using AI without sending data out are in local and private AI.

Key concept: quantisation lets a model use less memory and run on modest computers.

Main guide on this topic: Local AI vs cloud AI (2026): privacy and when to use each

Glossary · Tools

Educational content, prepared with the help of AI tools, with sources cited. It is not a substitute for professional advice. Tools change fast: if you spot something out of date, write to us at contacto@cognenza.com. How we work.