Your Mac Already Has a Free AI Model Built in: Here’s How to Unlock It with Apfel
TL;DR
Apple exposes its local Apple Intelligence model to developers through the Foundation Models framework. On compatible Macs, an roughly 3-billion-parameter model can run directly on the device. The open-source tool apfel adds three access paths: a CLI, a chat interface and a local server shaped around the OpenAI API format. The setup is aimed at small jobs such as shell commands, commit messages, short summaries, rewriting and tagging. In on-device mode it needs no API key, cloud bill or internet connection.
Nauti's Take
The headline sounds bigger than the product: your Mac has not suddenly gained a hidden frontier model. Still, apfel makes Apple’s developer angle clear: AI as a local operating-system capability, not another subscription tab with an API key.
That boundary is the useful part. Small, repeatable jobs can move onto the device; anything needing context, judgment and hard reliability still belongs with stronger models.
Briefingshow
This is not a new ChatGPT killer on the Mac, but a useful building block for private mini-automations. If a team handles short internal text, file names, tags or commit messages, local inference may be enough and sensitive data is more likely to stay on the device. It matters most for apps that want AI features without starting with a cloud dependency.
Video
Sources