microsoft/phicookbook

The Phi Cookbook is a collection of hands-on examples for working with Microsoft's Phi family of AI models. Phi models are small language models, meaning they are designed to run on modest hardware, including laptops and edge devices, rather than requiring large cloud servers. Despite their small size, the README describes strong performance on tasks like coding, reasoning, and text generation. The repository contains Jupyter notebooks, which are interactive documents that mix explanatory text with runnable code. You can open them directly in GitHub Codespaces (a cloud development environment that requires no local setup) or in a local development container. The notebooks walk through practical scenarios: running Phi models locally, connecting them to tools, building simple AI-powered applications, and working with images and audio in addition to text. The project is aimed at developers who want to experiment with generative AI without needing expensive infrastructure. Because Phi models can run on a phone or a laptop, the examples emphasize deployment to constrained environments as well as standard cloud setups. The README is available in many languages through automated translation, including Arabic, Chinese, French, German, Japanese, Korean, Spanish, and many others. Contributions are welcome. The project is maintained by Microsoft and participation in a community Discord server is encouraged. The full README is longer than what was shown.