Skip to content

Optional: Compare models

Time: 10 minutes Prerequisite: Complete the seven core steps first.

Step 1 introduced ModelSelector, which lists the models available to your signed-in account and stores the selected model ID in selectedModel. Every session configuration in the workshop uses that value through Model = selectedModel.

Run the completed application again:

Terminal window
dotnet run --project workshop-app

Choose a different number when the picker lists available models, then enter the workshop target URL. The application architecture, local tools, MCP configuration, and permission policy remain the same; only the model assigned to the session changes.

Model choice can affect latency, capability, and billing. It does not grant new tools or change the browser permission boundary. Keep tool allowlists and permission handling identical while you compare the response.

Troubleshooting this extension
SymptomFix
No models are listedThe helper falls back to the account default; verify authentication if this is unexpected.
A number is outside the rangeThe helper safely uses the first model.
Tools disappearKeep Model = selectedModel alongside the existing Tools, McpServers, and OnPermissionRequest settings.

The extension is complete when: you can name the selected model and explain why the same scoped tools and permissions still apply.

What does choosing a model change, and what does it leave unchanged?

Check your answer

It selects the model used by a CopilotSession, which can affect the response and latency. It does not change the client connection, application-owned tools, MCP server configuration, or permission boundary.

Return to Step 7: Run and explain the application.

Reset all workshop progress?

This clears completed lessons and checked task items stored in this browser. This action cannot be undone.