Free AI Software for Your Laptop Just Got a Small but Useful Fix

llama.cpp is free, open-source software that lets ordinary computers run AI chatbots themselves. Normally, using a powerful AI assistant means borrowing someone else's servers and paying a subscription every month. This project squeezes AI models down small enough that a laptop — or even a phone — can handle them. That means no monthly bill, and nothing you type ever leaves your device.

The change in this release is small and technical. It implements a "sameas" test for Jinja, the template language that AI models use to spell out how they want their instructions formatted. Think of it like a filling-in-the-blanks form. If the software misreads the form, the AI can ignore instructions, repeat itself, or give messy answers. This build makes that form-filling slightly more accurate and adds tests so it stays that way.

The bigger story is the list of devices attached to the release. There are versions for Apple Silicon Macs, Windows and Linux machines with Nvidia or AMD graphics cards, Android phones using Snapdragon chips, and even Chinese server processors. In practice, that means almost whatever device you already own can probably run an AI model locally. Two years ago, that was a hobbyist's dream. Now it's a routine download.

The catch: this is a pre-release build, one of thousands of numbered snapshots the project ships almost daily. It's a minor bug fix, not a flashy new feature, and installing it yourself still means using command-line tools. Most people should stick with friendly apps built on top of llama.cpp, such as Ollama or LM Studio. But every small fix like this quietly makes those easier apps work a little better for everyone else.