Free Tool That Runs AI on Your Own Laptop Just Got an Update

A free piece of software called llama.cpp released a small update on September 13. If you've never heard of it, here's the plain version: it lets you download an AI chatbot onto your own laptop or desktop and run it without sending your questions to a company's servers. No monthly fee. No internet connection required. No one reading your chats. The project has been starred (bookmarked) over 128,000 times on GitHub, which is a strong signal that a lot of people rely on it.

So what actually changed? Almost nothing you'd notice. The developers tweaked their internal testing so that one specific type of AI model, called HY_V4, is skipped when they check that the software works with WebGPU. WebGPU is a technical standard that lets a web browser or app use your computer's graphics card — the part normally used for video games — to speed up AI work. Think of it as rerouting traffic around a bumpy road so the rest of the cars keep moving.

Why should you care about something this small? Because llama.cpp is the engine behind a growing wave of free AI apps, and every little update keeps that engine running smoothly on more kinds of computers. The build listed support for Windows, Mac, Linux, Android and iPhone, and for graphics chips from Nvidia, AMD and Intel. That breadth is the point: the same free software works whether you have a gaming PC or a work laptop.

The catch is honest and worth stating. This release adds no new features you can use today — it's maintenance, the equivalent of an oil change rather than a new car. And running AI on your own machine still asks something of you: a reasonably modern computer, a few gigabytes of free storage, and a willingness to follow setup instructions. If that sounds like too much, the paid cloud chatbots remain far easier. But the gap is closing.