Copilot Usage Metrics API Adds OpenCode and More Tools
The Copilot usage metrics API has been updated to capture model usage from tools outside the traditional IDE, such as OpenCode, closing previous gaps in developer reporting.
Radwan AbdulhadiEditor and technology writer
Expanding Copilot Usage Metrics
GitHub has announced an update to the [GitHub Changelog](https://github.blog/changelog/2026-10-23-usage-metrics-api-adds-usage-from-tools-like-opencode/) regarding the Copilot usage metrics API. The system now reports model usage from a broader range of tools that developers use with Copilot, including tools like OpenCode that were previously uncaptured in reporting dashboards.
Because many of these tools operate outside the standard IDE and do not send editor telemetry, their activity was historically missing from totals_by_ide fields. Counting requests directly on the server helps close this measurement gap, allowing engineering teams to see which tools drive Copilot activity.
Understanding the New totals_by_client Array
The update introduces a new totals_by_client array available in both 1-day and 28-day reports across enterprise, organization, enterprise-user, and organization-user scopes. Each entry represents an individual client and reports key performance indicators.
The reported metrics include the specific client tool that transmitted the request, the count of successful model requests, distinct session counts, and detailed token usage covering prompt tokens, output tokens, and average tokens per request.
Tracked Client Tools and Identification
Recognized clients identified by the API include opencode, claude_code, codex, zed, copilot_cli, github_app, cloud_agent, code_review, vscode, and jetbrains. Any unrecognized clients are automatically grouped under other.
Clients are identified using the integration ID supplied with each incoming request. Consequently, the client parameter should be treated as a best-effort grouping rather than a fully verified identity.
Model Breakdown and Selection Modes
In addition to client-level tracking, the update provides a totals_by_model breakdown containing the same request and token metrics organized by model. This data includes a model_selection_mode field indicating whether a model was requested directly by the user, chosen automatically, or selected by the product.
For specialized features like Copilot code review and Copilot cloud agent, requests report product_selected because the system chooses the model automatically. Selection modes not yet reported individually appear as other, while unnormalized values are marked as unknown.
Distinction from Agent App Activity
GitHub notes that this client-based tracking is distinct from [agent app activity](https://github.blog/changelog/2026-08-06-copilot-usage-metrics-api-adds-agent-app-activity/), which counts specific jobs initiated by agent applications like Claude and Codex on GitHub.
By contrast, totals_by_client focuses purely on measuring model traffic—specifically requests and token consumption—from any tool leveraging Copilot as its underlying model provider.
Access Requirements and Documentation
Access to these updated metrics is restricted to enterprise owners, billing managers, organization owners, and users assigned a custom role containing the View Copilot Metrics permission. Additionally, the underlying Copilot usage metrics policy must be actively enabled.
Developers and administrators looking to incorporate these new fields into internal reporting workflows can consult the official [Copilot usage metrics API documentation](https://docs.github.com/rest/copilot/copilot-usage-metrics) for implementation guidelines.
Sources
- GitHub ChangelogUsage metrics API adds usage from tools like OpenCode
Continue chronologically
Related entity coverage
Related stories
Usage Metrics API Adds Pull Request Review Stages
Enterprise and organization repository-level Copilot usage metrics reports now break down how long pull requests spend in each review stage.
GitHub Updates Usage Metrics API with Custom Skills and MCP Servers
GitHub updates the Copilot usage metrics API to display specific names for custom skills, agents, MCP servers, slash commands, and plugins.
GitHub Copilot Expands Code Review Configurations
GitHub Copilot code review introduces general availability for personal configurations across more plans and adds enterprise-level default settings.
GitHub Announces Important Updates and Bug Fixes for Secret Scanning
GitHub has rolled out a series of updates, improvements, and bug fixes for secret scanning, including changes to audit log events and custom patterns.
