Free Update Makes AI Run Faster on New Macs

Imagine running a smart AI assistant entirely on your laptop — no internet, no sending your messages to a company server. That's what llama.cpp does. It's a free, open-source program that lets people run AI models like ChatGPT-style bots on their own computer. Think of it as a home engine that many AI apps secretly use under the hood.

This week, the project released a new pre-release version called b10770. The big change is a speed boost for Apple's M3 chip, the processor in the latest MacBooks and iMacs. The update fine-tunes how the software talks to the M3's specialized hardware, so AI text can generate faster. In plain English: your on-device AI will feel less like waiting for a slow typist and more like chatting with a quick-thinking friend.

Why should you care if you're not a developer? Because many apps quietly rely on llama.cpp. That includes local AI writing tools, coding helpers, and privacy-focused chat apps. When this software improves, those apps improve too. And because everything is processed directly on your Mac, your private conversations never leave your computer. No cloud, no prying eyes — just you and your data.

Of course, there's a catch. This is an early testing release, so it's not polished for everyday users yet. If you own an M3 Mac, you can try it now, but most people should wait for the stable version. Still, this update signals that on-device AI is getting faster. That's good news for anyone who wants powerful AI without sacrificing privacy or wallet-draining subscriptions.