llama.cpp Update Keeps Free AI Running Privately on Your Laptop

On September 13, an open-source project called llama.cpp released a new version, labeled b10935. If you've never heard of it, here's the short version: llama.cpp is the free engine that lets you download an AI chatbot — similar to the ones from OpenAI or Google — and run it on your own laptop or phone. No monthly fee. No internet needed. Your questions never leave your device. The project has 128,000 stars on GitHub, making it one of the most popular pieces of AI software in the world.

The specific change in this release is tiny and technical: a new "LOG_JSON" tool that lets the program record what it's doing in a structured format other software can read. Think of it like a delivery driver switching from scribbled paper notes to barcode scans. It doesn't make the AI smarter. It makes it easier for developers to spot bugs and fix them faster, which eventually means fewer crashes for everyone.

The more interesting part is the sheer list of things this release supports: Apple Silicon Macs, iPhones, Windows PCs, Android phones, Linux servers, and graphics cards from Nvidia, AMD, and Intel. That breadth is the real point of llama.cpp. It's the reason the same AI model can run on a gaming PC, a MacBook, or a mid-range phone. Competing tools tend to lock you into one company's hardware.

Now the honest catch. This is a pre-release developer build — plumbing, not a product. You can't just download it and start chatting. If you want to try local AI today, free apps like LM Studio or Ollama wrap this engine in a friendly interface. And remember the trade-off: local models are still smaller and less capable than ChatGPT or Gemini. You're buying privacy and zero cost with a bit of brainpower.