Rhea Patel and Samruddhi Khandale introduce a new Visual Studio Chat feature allowing developers to bring their own language models from providers like OpenAI, Anthropic, and Google, enabling customization, flexibility, and direct API integration.

Bring Your Own Language Model to Visual Studio Chat

Authors: Rhea Patel, Samruddhi Khandale

Developers working in Visual Studio Chat can now connect their own large language models by providing API keys from providers such as OpenAI, Anthropic, and Google. This enhancement allows users to:

Select from a wider range of language models by integrating third-party or latest releases immediately
Customize workflows to match organizational security, infrastructure, or performance needs
Directly manage usage, quotas, and billing with chosen providers
Switch between Copilot’s built-in models and external models for enhanced flexibility

How to Enable

Open the Chat Window in Visual Studio.
Click on Manage Models from the model picker.
Choose your preferred provider and enter your API key.
Select from preset models or enter a custom model name.
The connected model will appear in your model picker for use in chat.

Key Considerations

Feature Scope: This applies strictly to Chat within Visual Studio, not completions, commit messages, or other AI features.
Model Capability Differences: Not all external models offer features like tool use or vision support.
Copilot Business/Enterprise Restriction: This integration is unavailable for Copilot Business and Copilot Enterprise users.
Partial Integration: Embeddings, repo indexing, and intent detection still use Copilot APIs.
Responsible AI: Output from third-party models may bypass Copilot’s responsible AI filters—use with awareness.

Further support for additional AI providers is anticipated. For more details or to try it out, visit Visual Studio.

This post appeared first on “Microsoft DevBlog”. Read the entire article here