The Free App That Runs ChatGPT-Style AI on Your Own Laptop

There's a free app called llama.cpp, and it does something the big AI companies would rather you didn't know was possible: it runs powerful AI chatbots entirely on your own computer. No cloud, no subscription, no internet connection required. This week the project released a new build (they ship updates constantly — this one is numbered b11022), and it has been downloaded and starred by more than 129,000 people on GitHub, the website where programmers share code.

So what actually changed? Mostly plumbing. The update reorganizes how the app talks to Vulkan — that's the technology that lets AI use your computer's graphics card instead of just its main processor. Think of it like rewiring a kitchen so the stove and the fridge stop sharing one outlet. Nothing new appears on your plate, but everything runs more smoothly, and it's easier for other programmers to build on top of it. The release also bundles ready-to-run versions for Windows, Mac, Linux, Android and iPhones, and for every major graphics chip brand — Nvidia, AMD, Intel and others.

Why should you care? Three reasons. First, privacy: when the AI runs on your machine, your questions, documents and late-night worries never travel to a company's servers. Second, money: there's no $20-a-month bill. Third, staying power: it works on a plane, in a basement, or during an outage.

The catch is real, though. You need a reasonably modern computer with a decent graphics card to run the good models, and you have to download the AI model file yourself — it can be many gigabytes. Setting it up is still a tinkerer's job, not a one-click install. But the direction is clear: high-quality AI is slowly becoming something you own rather than something you rent.