Your Actions allowance is a shared wallet now - and your agent is spending from it #213355
Replies: 1 comment
|
Thank you for your interest in contributing to our community! We currently only accept discussions created through the GitHub UI using our provided discussion templates. Please re-submit your discussion by navigating to the appropriate category and using the template provided. This discussion has been closed because it was not submitted through the expected format. If you believe this was a mistake, please reach out to the maintainers. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Worth re-reading the Actions billing page if you build with agents, because the shape of the meter changed and most "CI is expensive" conversations haven't caught up.
Included compute usage is no longer a minutes quota. It's a dollar amount — $12 on Free, $18 on Pro/Team, $300 on Enterprise Cloud — and several things draw from it: standard hosted runners, cloud sandbox compute and memory, larger runners on Team/Enterprise, and Copilot code review on private repos. The docs are blunt about the consequence: no eligible usage type has priority over another when drawing from the shared balance.
Which means a review bot someone enabled six months ago and a sandbox an agent spun up at 2am are now competing with your test suite for the same money. And when it runs out, nothing degrades gracefully. Jobs die at setup in a few seconds, which looks exactly like a broken pipeline until you notice the meter is empty.
The per-minute rate isn't the lever anyway. Rates are genuinely cheap now — $0.006 for a 2-core Linux runner, $0.062 for macOS. What compounds is job count. Every job pays its own start and setup, partial minutes round up per job, and splitting a slow suite into twelve shards makes the invoice grow while the wall clock shrinks. One team published a month of 9,700 workflows that worked out to roughly $9 of actual compute, with plan limits on job starts pushing it past $500.
So here's the part I actually want to argue about. When agents multiply your PR volume, do you cut how often you ask CI a question, or do you move the gate somewhere cheaper entirely? Self-hosted runners are still listed as free, which sounds like the answer until you price the maintenance and the "works on the runner, not on your laptop" class of bug.
What does the line item actually look like on your account — and did anyone consciously agree to it, or did it just grow?
All reactions