GitHub has updated its Copilot usage report to include a detailed, per-model breakdown of token consumption. This enhancement allows users to see the specific input, output, cache read, and cache write tokens associated with their AI credits for each model.
Previously, the report displayed only AI credits without underlying token details, making it challenging to understand or optimize costs. The new breakdown aims to provide transparency into how token usage contributes to each model's cost.
Key Points
- The Copilot usage report now shows a per-model breakdown of tokens.
- For each model, the report details input, output, cache read, and cache write tokens.
- The report also displays the AI credits consumed by each model.
- This token breakdown is available to admins on Copilot Business and Copilot Enterprise.
- Individuals using Copilot for individuals also have access to this feature.
- Users can download the AI usage report from the AI usage page within their billing settings.
Context
According to GitHub, the previous usage report showed AI credits without the underlying token detail, which complicated efforts to explain charges or identify cost-reduction opportunities. The new token breakdown allows users to trace how various token types contribute to each model's cost and share these details with stakeholders.
Why It Matters
This update provides greater transparency into AI credit consumption for Copilot users, enabling better cost management and clearer communication regarding AI resource utilization.
What To Do
- Download the AI usage report from the AI usage page in your billing settings.
- Review the per-model token breakdown to understand input, output, and cached token usage.
- Compare token consumption across different models to identify cost drivers.
- Share the detailed token breakdown with stakeholders to explain AI credit charges.