Allows using local Llama.cpp models within GitHub Copilot for VS Code. Saves costs and provides more control over the models used in Copilot Chat, enabling private and customizable AI assistance for developers.