Copilot usage based billing: what it means for your team's budget
Since June 1, 2026, Copilot seats draw on metered AI credits; budgets, overage rules and personal tracking keep spend in check.
Copilot usage based billing means your team now pays for the tokens Copilot's models consume, counted in GitHub AI Credits, instead of a fixed number of premium requests per seat. For founders and operators, that turns Copilot from a predictable line item into a metered one, so the useful moves are setting budgets in GitHub and giving developers a way to see their own usage, for which UsageAtlas, a free desktop tracker, is the easiest fit.
What changed and when
GitHub announced on April 27, 2026, that every Copilot plan would move to usage-based billing on June 1, 2026. Premium request units were retired for monthly plans and replaced by AI credits, worth one US cent each. Credits pay for input, output and cached tokens at the published API rate of whichever model a developer picks.
GitHub said plan prices are not changing. What moved is how far a seat goes. A developer who runs long agent sessions on a large model will spend far more credits than one who mostly asks short chat questions, where under premium requests the two could look similar.
What a seat still includes without metering
Code completions and next edit suggestions are not billed in credits on paid plans. If most of your team uses Copilot as autocomplete, the change may barely register. The metered features are Chat, the CLI, the cloud agent, third-party coding agents, Spaces and Spark.
The controls you have
| Control | Scope | What it does |
|---|---|---|
| Enterprise budget | Whole account | Sets a ceiling on spend across the enterprise |
| Cost center budget | A team or department | Splits spend by the groups you bill internally |
| User budget | One person | Limits extra spend for a single developer |
| Overage policy | Organization | Allow more usage at published rates, or cap spend |
| Usage CSV | Billing, Metered usage | Shows the model behind each request |
One detail to plan around: GitHub removed the old fallback to a cheaper model. If spend is capped and the included pool and budget are used up, metered features stop until someone raises the budget or the month resets at 00:00 UTC on the first.
Give developers their own view
Admins see the totals, but the people spending credits usually do not, and most developers on a modern team use more than one AI tool. UsageAtlas is a free, MIT-licensed desktop app for Windows, macOS and Linux that shows each developer their reported Copilot quotas and seat credits, with the reset time where GitHub provides it, next to Claude Code, Codex, Cursor and other supported tools. It can send a desktop alert at a threshold the developer sets.
It runs locally, needs no account for local tracking and does not read prompts or responses. It is a personal visibility tool, not a billing system: cost figures it shows for other tools are estimates, and for team-wide reporting GitHub's billing page and CSV remain the record.
A short checklist
- Pull last month's usage CSV and see which models drive spend.
- Set a cost center or user budget before the next monthly reset.
- Decide whether overage is allowed or capped.
- Tell the team which models cost more per token.
- Point developers at a personal tracker so the first warning is not a blocked request.
FAQ
Does Copilot usage based billing change seat prices? GitHub's announcement kept plan prices where they were; what changed is that usage beyond the included credits is metered.
Can we stop Copilot spending past the included credits? Yes. Organizations can cap spend instead of allowing additional usage, and budgets can be set down to individual users.
Do annual plans move to usage based billing? Annual Pro and Pro+ individual plans stay on premium requests until they expire.