GitLab CI compute minutes used up: "No more compute minutes available"
Short answer
On GitLab.com, jobs on GitLab-hosted runners use compute minutes from the top-level group's or personal namespace's monthly quota: 400 on Free, 10,000 on Premium and 50,000 on Ultimate. A job uses its running time multiplied by the runner's cost factor, 1 for the default small Linux runner, more for bigger, macOS and GPU runners. When the quota runs out, hosted runners take no new jobs, and jobs fail with "No more compute minutes available". Usage resets on the first of each month. To keep going sooner, buy more minutes, at $10 per 1,000, or run jobs on your own project or group runner, which uses no quota.
Pipelines that worked yesterday now fail before their jobs start, and each job says:
No more compute minutes available
On GitLab.com, jobs on GitLab-hosted runners use compute minutes, and your namespace’s monthly quota is used up. Here’s how the quota works in GitLab 19.5, and the ways to keep pipelines running.
How the quota works
- One quota per top-level namespace. Every project in a group and its subgroups shares the group’s quota; personal projects share yours. A pipeline in someone else’s group uses that group’s minutes, not yours.
- The monthly amount depends on the plan: 400 compute minutes on Free, 10,000 on Premium, 50,000 on Ultimate.
- Usage resets to 0 on the first day of each month.
- Only running time counts. A job uses its running time multiplied by the runner’s cost factor. Time spent pending doesn’t count, and trigger jobs use nothing.
Cost factors
Untagged jobs run on the small Linux runner, with a cost factor of 1. Larger runners use minutes faster:
| Hosted runner | Cost factor | Plans |
|---|---|---|
| Linux small, x86-64 or Arm64 | 1 | All |
| Linux medium | 2 | All on x86-64; Premium and Ultimate on Arm64 |
| Linux large | 3 | Premium, Ultimate |
| Linux xlarge | 6 | Premium, Ultimate |
| Linux 2xlarge | 12 | Premium, Ultimate |
| Linux medium with GPU | 7 | Premium, Ultimate |
| macOS M1 medium | 6 | Premium, Ultimate |
| macOS M2 Pro large | 12 | Premium, Ultimate |
| Windows medium | 1 | All |
A 10-minute job on a medium Linux runner uses 20 minutes, so a Free namespace gets about 200 minutes of medium-runner time a month. The most common reason to pay for a bigger runner is a job that runs out of memory on the small one: see GitLab CI exit code 137. Arm64 runners need images built for arm64, or jobs fail with “exec format error”: see GitLab CI “exec format error”. Public projects in the GitLab for Open Source program have a cost factor of 0.5.
Check your usage
- A group: with the Owner role, open the top-level group, then Settings › Usage quotas, and select Pipelines. It lists this month’s projects by usage.
- Your personal namespace: select your avatar, then Edit profile, then Usage quotas.
Owners get a banner and an email when less than 25% is left, less than 5% is left, and when it’s used up.
What happens when it runs out
- GitLab-hosted runners take no new jobs. Pending jobs, and jobs you retry, fail with “No more compute minutes available”.
- Running jobs continue until the namespace is 1,000 minutes over its quota, then they’re stopped too.
- Your own runners keep working. Project and group runners aren’t limited by the quota.
Jobs that just wait can look like a runner problem. If they’re pending rather than failing, see jobs stuck pending.
Keep pipelines running
- Wait for the 1st. The quota refills at the start of the month.
- Buy more minutes with Buy additional compute minutes on the Usage quotas page: a group’s Owner or billing account manager for a group, or you for your personal namespace. They cost $10 per 1,000 minutes, are used only after the monthly quota, roll over to the next month, and are valid for 12 months.
- Run jobs on your own runner. With the Maintainer role, go to Settings › CI/CD, expand Runners, select Create project runner, select Run untagged unless your jobs have tags, and register it on a machine you control. Then turn off Turn on instance runners for this project in the same section, so jobs don’t fall back to GitLab’s runners.
- Upgrade to Premium or Ultimate for a larger monthly quota.
Use fewer minutes
- Skip pipelines you don’t need, such as for documentation changes: see how to skip a GitLab CI pipeline.
- Run jobs only when they matter, with
rules, and avoid duplicate pipelines for the same push. - Mark jobs
interruptible, so a newer pipeline on the branch cancels the older one. - Run schedules less often, and turn off pipelines for pull mirror updates if you mirror a repository.
- Use the cache for dependencies instead of downloading them in every job: see GitLab CI cache not working.
A different message: identity verification
On GitLab.com Free, a job on GitLab-hosted runners can fail with “Identity verification is required in order to run CI jobs” even with minutes left. That’s an anti-abuse check, based on the account’s risk score: select Verify my account in the banner, verify your phone number or add a payment method if asked, then start a new pipeline.
On your phone
MergeBell shows a pipeline’s jobs and their status at a glance, so a run of jobs that failed before starting stands out, and Retry reruns a job on the free tier once minutes are back or a runner of your own is online. Usage quotas and buying minutes need GitLab in a browser.
Frequently asked questions
Why do my GitLab jobs fail with "No more compute minutes available"?
Your top-level group or personal namespace on GitLab.com has used its monthly compute quota, so GitLab-hosted runners stop taking its jobs. Pending and retried jobs fail with this message, and running ones are stopped once usage is 1,000 minutes over. Wait for the reset on the first of the month, buy more minutes, or run the jobs on your own runner.
How many compute minutes does GitLab.com give for free?
400 compute minutes a month for a Free namespace, counted for the whole top-level group or personal namespace, not per project. Premium includes 10,000 and Ultimate 50,000. A job on the default small Linux runner uses one minute per minute it runs; larger runners use more.
How are GitLab compute minutes calculated?
Each job uses its running time in minutes multiplied by the runner's cost factor: 1 for small Linux runners, 2 for medium, 3 for large, 6 for xlarge, 12 for 2xlarge, 7 for the GPU runner, 6 for macOS M1 and 12 for macOS M2 Pro. Time spent pending doesn't count. Jobs that run in parallel all count, so a pipeline can use more minutes than its wall-clock time.
Where can I see my GitLab compute minutes usage?
For a group, its Owner opens the top-level group's Settings › Usage quotas › Pipelines. For your personal namespace, select your avatar, then Edit profile › Usage quotas. Both show the minutes used this month and the projects that used them.
Do my own GitLab runners use compute minutes?
No. The quota applies only to GitLab-hosted instance runners. Jobs that run on a project or group runner you host yourself keep running when the quota is used up. Turn off instance runners for the project so untagged jobs don't go to GitLab's runners instead.
Sources
- GitLab docs — Compute minutes
- GitLab docs — Compute usage for instance runners
- GitLab docs — Purchase additional compute minutes
- GitLab pricing — compute minutes per tier
- GitLab docs — Create a project runner
- GitLab docs — Identity verification is required in order to run CI jobs
Checked against these pages on 7 November 2026. GitLab changes quickly — if something here no longer matches, tell us and we will fix it.
Do this from your phone with MergeBell
Free for the whole review loop: the To-Do inbox, diffs and line comments, approve and merge, pipelines with retry, and push alerts for your To-Dos. Works with gitlab.com and self-managed GitLab 16.0+.
Get MergeBell for iPhoneGet MergeBell for Android
On the App Store for iOS 26 or later and on Google Play for Android 12 or later. An independent app, not affiliated with GitLab Inc.