GitHub Copilot token billing drains AI credits, users report

GitHub changed Copilot billing in April 2026 to charge by token consumption. Since the announcement, developers have posted reports on GitHub Community Discussions that monthly AI credit balances are falling faster than expected.

Users shared specific examples of rapid credit use. One forum poster, rvs99, wrote: “My 12% of total AI credits burned like anything for very minor task. I used Claude Sonnet 4.6 as usual and in response it barely updated 2-3 lines in total 6 files which costed like ~$0.35 per line updates.” Another participant, prhost, posted a dashboard screenshot showing 3,705 credits remaining from an allowance of 7,000 after one day and added: “It would be easier to shut down the project. [Microsoft] shot themselves in the foot.” A third poster, zoomp05, commented: “The strategy is clear, but it would have been good to say from the beginning, ‘This is a subsidized trial’ or something similar, to promote our tool.”

Before April, subscription plans let users access AI features up to fixed limits without direct per-token charges; those plans are now deprecated. Users report that calls to larger or higher-rate models consume more tokens and therefore more of their monthly credits.

Operating large language models involves costs for model development, post-training work, upkeep, data center operation and related infrastructure. Charging by token makes costs align more closely with how much compute and capacity a given request uses.

The impact varies by user. Some report only modest increases in monthly spending, while others say routine code edits now use substantial shares of their credits. Forum posts concentrate on immediate billing effects and requests for clearer information about how token consumption maps to credit depletion.