How to run AI locally on your own computer: best free apps
Updated · Prices change often; confirm on each tool's site.
The easiest way to run AI on your own computer is LM Studio, a free desktop app where you download an open model and chat with it offline. Ollama is the favorite for developers who want a one-command setup and a local API, and Jan is a fully open-source ChatGPT-style app. All three run on Windows, macOS and Linux and are free for local use.
Want AI that works offline, keeps your data on your machine and has no monthly bill? Free apps now make running open models on a normal computer a few clicks.
Our top 3 picks
LM Studio
Point and click: search for a model, download it and chat, with an OpenAI-compatible local server when you need one. Free for home and work use; optional paid cloud plans.
Ollama
One command to download and run open models, plus a local API that many coding tools and apps plug into. MIT-licensed and free locally; optional Ollama Cloud from $20/mo.
Jan
Open-source (Apache 2.0) ChatGPT alternative that runs models fully offline, with an optional local API and cloud model connections if you want them. Free.
How to do it
- Check your hardware: more RAM (and GPU memory, or an Apple Silicon Mac's unified memory) lets you run bigger, smarter models. Small models run on most recent laptops.
- Install LM Studio, Ollama or Jan.
- Download a small model first (a few billion parameters, in a 'quantized' version) and move up only if it runs smoothly.
- Chat offline, or turn on the local server to connect your own apps or a code editor.
- Check each model's license before using it in a commercial product.
FAQ
Is running AI locally free?
Yes. LM Studio, Ollama and Jan are free to run on your own computer; you pay only for the hardware and electricity. Their optional cloud plans are separate and not needed for local use.
What computer do I need?
Small models run on most modern laptops, though slowly on older ones. For bigger models, 16 GB of RAM or more and either a recent NVIDIA/AMD graphics card or an Apple Silicon Mac is a sensible start. A model's download size is a rough guide to the memory it needs.
Is local AI as good as ChatGPT or Claude?
Usually not on the hardest tasks, because the biggest frontier models don't fit on a laptop. Local models are good for private notes, drafting, summarizing and coding help, and they work offline.
Is local AI private?
With a local model, your prompts and files stay on your computer. If you turn on cloud models or web search inside these apps, that data goes to the provider.
Ollama or LM Studio?
LM Studio if you want a visual app; Ollama if you're comfortable in a terminal or want to plug models into other tools. Many people install both.
Related guides
How we pick
We match tools to the task, not to commissions. Several of the best tools we recommend (e.g. ChatGPT, Claude, Midjourney, Cursor) pay us nothing. Picks come from a curated, hand-checked list and are refreshed regularly. See our method and affiliate disclosure.
Got a more specific task? Try your own task.
Describe it and get matched →