Windows-Copilot-API

Why we included this project

This project fills a simple need: a free way to call a strong model from your own code. It wraps the free Microsoft Copilot chat in an OpenAI-compatible REST server, so the official openai SDK and most OpenAI-based tools can point at http://localhost:8000/v1 instead of OpenAI's endpoint with minimal changes. You sign in once in a browser with a Microsoft or Google account, and the session is saved and refreshed automatically. It streams tokens and supports multi-turn threads via conversation_id, and a bundled Python client covers the case where you'd rather call client.chat() directly than run the server. Just know it's an unofficial automation of the consumer web experience: expect rate limits and Cloudflare checks, and the signed-in path only works in regions where signed-in Copilot is available. For personal projects that need a no-cost model endpoint, it's a practical shortcut.