BurntAI

How to track GitHub Copilot and Claude spend per developer

By Jiri Sifalda

To track GitHub Copilot and Claude spend per developer, first find out how each developer reaches each tool. A developer can hold a Copilot seat, a Claude Team or Enterprise seat, and an Anthropic API key at the same time. Each path has its own report, its own identity and its own delay. No single report from GitHub or Anthropic covers all three.

This guide lists the paths, names the per-person source for each, and shows how to add them into one monthly number per developer.

Map the paths before you pull any data

Make a list with one row per developer and one column per path:

  • Copilot seat. A Business or Enterprise seat in your GitHub organization.
  • Claude seat. A Standard or Premium seat on a Claude Team or Enterprise plan. Claude Code is included in both Team seat types (Claude pricing).
  • Claude API. Claude Code or other tools that sign in to your Claude Console organization with an API key or a Console account.

A developer with a Claude seat and an API key appears in two different Anthropic reports. If you read only one, you miss part of their cost. If you add the seat and the API usage without a plan, you can also count the same person twice under two identities.

The Copilot side: one source per person

Copilot is the simpler half. Each person costs their seat price plus the AI credits they consumed. GitHub lists Copilot Business at $19 and Copilot Enterprise at $39 per granted seat per month (GitHub Copilot plans), and 1 AI credit = $0.01 USD (usage-based billing for organizations and enterprises).

For the credits, use the ai_credits_used field in the user-level Copilot usage metrics reports (GitHub changelog), or the emailed AI usage report for a monthly review. Data for one day is available within two full UTC days after that day closes (About Copilot usage metrics). The Copilot cost per user guide explains each report and its limits.

The Claude side: the source depends on the path

Claude API users in a Console organization

For a Claude Console organization, you have two per-person sources:

  • Claude Code Analytics API. It returns daily Claude Code metrics per user, with an estimated_cost per model (Claude Code Analytics API). It needs an Admin API key from the Claude Console.
  • Usage and Cost Analytics API. Call GET /v1/organizations/analytics/cost with group_by[]=actor to get cost per user. Add products[]=claude_code to limit it to Claude Code (Usage and Cost Analytics API).

The cost endpoint returns two amounts. amount is the cost after discounts and before any credits are applied. list_amount is the cost at list prices. Use amount to match what you pay, and list_amount to compare developers when one team has a discount and another does not.

Claude Enterprise seats

A Claude Enterprise organization uses the Claude Enterprise Analytics API. Only the primary owner can create its Analytics API key, in claude.ai organization settings (Analytics APIs). The same Usage and Cost Analytics API endpoints serve Enterprise organizations, so you can use the same group_by[]=actor call.

Read the scope note on that page with care. The cost and usage reports apply to usage-based Enterprise plans. For seat-based Enterprise plans, they reflect usage credits only. On a seat-based plan, the per-person cost is the seat fee plus the credits the report shows.

Claude Team seats

The Anthropic API docs above list Console and Enterprise organizations. They do not list the Team plan. On a Team plan, the documented per-member view is in the admin settings. When members go past their included usage, they can continue on usage credits, which are billed at standard API rates. Owners can see each member's spend in the MTD Spend column under Spend limits by user in Organization settings > Usage (manage usage credits for Team and seat-based Enterprise plans).

So a Team developer costs their seat price, $20 per month for a Standard seat billed annually or $100 for a Premium seat (Claude pricing), plus their month-to-date usage credit spend from that column.

A fallback for every path: OpenTelemetry

Claude Code can export its own metrics. Set CLAUDE_CODE_ENABLE_TELEMETRY=1 and choose an exporter, and Claude Code emits claude_code.cost.usage in USD, tagged with user.email (Claude Code monitoring). Administrators can set this for all users through the managed settings file.

This works the same for seat users and API users, so it is a good cross-check. It is not a bill. The docs state that cost metrics are approximations, and a seat user's cost inside their included usage is not an extra charge.

Know the delay of each source

A per-developer total is only as current as its slowest source. Mark the last complete day for each one:

  • Copilot usage metrics: up to two full UTC days after the day closes.
  • Claude cost and usage reports: typically within 4 hours, up to 24 hours. Values can change until about 7 days after the end of the calendar month (Analytics APIs).
  • Claude Team MTD Spend column: a view in the admin settings, not a feed. Someone must read or export it on a fixed day.

For a month-end review, wait until about the 8th of the next month before you treat the Claude numbers as final.

Add it up per developer

  1. Pick one key per person. GitHub reports a login. Anthropic reports an email address or a user ID. Keep one mapping table from GitHub login to work email.
  2. Add the fixed part. For each person, add their Copilot seat price and their Claude seat price, if they hold one.
  3. Add the metered part. Add Copilot credits times $0.01, Claude API cost from the cost endpoint, and Claude usage credits from the Enterprise report or the Team MTD Spend column.
  4. Do not count one usage twice. If you also collect OpenTelemetry cost, use it as a check, not as a separate line.
  5. Compare to a budget. Put each person's month-to-date total next to their budget and the share of the month that has passed.
  6. Reconcile monthly. Compare the sum of all developers against the GitHub and Anthropic invoices, and record the difference.

The per-developer AI spend tracker guide shows how to store these rows day by day.

Let one tool do the joining

BurntAI is built for this exact join: one colleague's Copilot seat, Anthropic API key and Claude seat on one row, with seat fees included, against a monthly budget and a month-end forecast. See pricing to get on the list.

See every dollar your team spends on AI

See every developer's GitHub Copilot and Claude spend in one dashboard. Track per-person AI costs, set budgets, and catch overspend before the invoice lands.