Free App That Runs AI on Your Own Laptop Gets a Fix

Meet llama.cpp. It's free, open-source software — built by volunteers and hobbyists, now starred by more than 130,000 people on GitHub — that lets you run an AI chatbot on your own laptop, desktop, or phone. Think of it as the difference between cooking dinner at home and ordering takeout every night: instead of paying a company monthly to use their AI, you run the AI yourself, offline, using your own hardware.

What actually changed in this update, labeled b11312, is tiny. It fixes a technical bug where two parts of the program were accidentally sharing the same spot in memory while running a certain kind of AI model inside a web browser. In plain terms: a mix-up that could cause garbled or broken results got cleaned up. If you're not a programmer, you'd never notice the difference — and that's the point. Someone did notice, reported it, and it got fixed within days.

The bigger story is what this project quietly represents. Every build listed in the release works on something different: Macs with Apple chips, ordinary Windows PCs, Linux servers, Android phones, even Snapdragon chips and Chinese-made hardware. That means the same free AI software runs on the phone in your pocket and the computer on your desk. Because nothing is sent to a company server, your notes, medical questions, or work documents stay on your device.

The honest catch: this is a pre-release, which means it's a work-in-progress build rather than a polished consumer app. You have to install it yourself, and the AI models small enough to run on a personal computer are generally less capable than the giant ones behind ChatGPT or Gemini. So this isn't a replacement for those services — it's an alternative for people who value privacy, hate subscriptions, or want AI that works with no internet at all.