Free Tool Lets You Run ChatGPT-Style AI on Your Own Laptop

A free, open-source project called llama.cpp — think of it as a homemade alternative to ChatGPT that runs entirely on your own hardware — published a new version on September 16. There is no company charging you and no website storing your chats. Once it's on your machine, the AI answers offline, and everything you type stays on your device.

The main thing that arrived is a fresh batch of ready-made installers. Instead of making you build the software yourself, the team now ships prepackaged downloads for Macs with Apple chips, Windows PCs (including gaming graphics cards from Nvidia and AMD), Linux servers, Android phones, and even specialised Chinese chips. That matters because installing this kind of software used to take an afternoon and some coding knowledge. Now you mostly download a file and run it.

So why should you care? Two reasons: money and privacy. Cloud AI subscriptions cost roughly $20 a month, and everything you type goes to someone else's computers. Running a model locally is free after setup, works on a plane, and keeps your medical questions, work documents, or embarrassing search history to yourself. You do need a reasonably modern computer; a laptop from the last four or five years can handle smaller models, and a gaming PC with a dedicated graphics card handles bigger, smarter ones.

The honest catch is that this particular update is minor. The one real change is a small utility fix that prints version information — useful for developers debugging their setup, invisible for everyone else. The friendlier installers are the substance here, not new brainpower. And while the downloads are easier, this is still a tinkerer's tool. If you want AI that just works with a chat box and a login, a paid service is still the smoother path.