Feature request: Implement GH Copilot Language Model Chat Provider

[prawel]

In the newest (September 11th) release of VS Code, there is new API present: Language Model Chat Provider API. It enables use of custom models in GH Copilot extension.

Official link:
https://code.visualstudio.com/api/extension-guides/ai/language-model-chat-provider

Is this something that you consider in llama.vscode?

Cheers!

[igardev]

Thank you for the idea. I will have it in mind. However, there are two reasons, which make me give it a lower priority. First, VS Code Copilot will add a new feature, which will allow a running llama-server to be used directly. This way it will be possible to find/download/start models with llama-vscode and use the endpoints from VS Code with Copilot even without using Language Model Chat Provider API. The second reason is that "Models provided through this API are currently only available to users on individual GitHub Copilot plans". I prefer focusing on features, which do not require Copilot plan.

[prawel]

Sure, that focus makes sense.

Just to clarify, I think "individual GH Copilot plans" also includes free one, and at least free one is required anyway for using custom OpenAI endpoints they plan to add in the next release.

Language Model Chat Provider API will allow us to use built in Copilot Chat tooling, and better VS Code integration.