Better Visibility Into Model Usage

GitHub Copilot's usage metrics now provide complete clarity on model consumption when auto model selection is enabled. Previously, activity that fell under auto mode appeared as a generic "Auto" label, making it difficult for organizations to understand their actual model usage patterns. This update resolves that data to show the specific model names being used.

What's Changed

The improvement applies across multiple reporting surfaces:

• REST API: The totals_by_model_feature endpoint now resolves auto selections to actual models across enterprise, organization, and user-level reports
• Dashboard: "Model usage per chat mode" charts now display granular model breakdowns instead of aggregated auto metrics

Auto usage is automatically rolled into the resolved model totals, providing a unified view without requiring separate tracking of auto vs. manually selected model usage.

Why This Matters for Admins

• Compliance and auditing: Organizations can now accurately track which specific models power their Copilot deployments
• Cost optimization: Granular model-level metrics enable better understanding of resource allocation and usage patterns
• Operational clarity: As auto mode adoption grows, admins maintain clear, actionable visibility into their fleet

Note that a separate breakdown specifically comparing auto vs. manually selected model usage is not yet available but is being considered for future updates.