Bring Your Own Model to the GitHub Copilot SDK with C#

Microsoft introduced bring-your-own-model support for the GitHub Copilot SDK using C# and .NET 10. The capability allows developers to combine the Copilot agent runtime with external model providers, routing inference through a configuration object called ProviderConfig.

Supported providers include OpenAI, Anthropic, Ollama, Microsoft Foundry Local, and OpenAI-compatible endpoints such as vLLM or LiteLLM. A sample console application published in the Generative AI for Beginners .NET repository demonstrates how to configure a Microsoft Foundry deployment with the runtime.

The sample project uses GitHub.Copilot.SDK alongside Azure.Identity to authenticate via Microsoft Entra ID or an API key. Configuration details specify model names, endpoints, and token limits, including MaxPromptTokens set to 8000 and MaxOutputTokens capped at 512. Authentication defaults to reusing credentials from az login via AzureCliCredential.