Docs

Runners and sizes

Every job gets its own fresh machine, started for that job and gone when it ends.

Labels

Pick a size with the job’s runs-on label. checklane on its own is the 4 vCPU machine.

Label	vCPUs	Memory
`checklane`	4	16 GB
`checklane-2vcpu`	2	8 GB
`checklane-4vcpu`	4	16 GB
`checklane-8vcpu`	8	32 GB
`checklane-16vcpu`	16	64 GB

jobs:
  build:
    runs-on: checklane-8vcpu

Your plan sets the biggest machine a job gets: 4 vCPUs on Pay as you go, 8 on Pro and 16 on Team. A job that asks for more runs on the biggest your plan has, so it never gets stuck. See Pricing.

What’s on the machine

Runners start from Ubuntu 24.04 with what a workflow written for ubuntu-latest expects to find: Git, Docker with Buildx and Compose, the GitHub CLI, jq, yq and the usual build tools. Node.js 20, 22 and 24, Python 3.11 to 3.13 and Go sit in the tool cache, so actions/setup-node, actions/setup-python and actions/setup-go find them without downloading.

Anything else installs the way it does on GitHub’s runners, with a setup action or sudo apt-get install. The tool cache is kept between runs, so what a setup action downloads once is there next time.

Where jobs run

Jobs run on our own servers. When they’re busy, a job can run on a partner’s machine of the same size instead of waiting. The label, the image and the price stay the same either way.

Jobs at once

Your plan sets how many vCPUs your jobs use at the same time: 4 on Pay as you go, 8 on Pro, 16 on Team. When you’re at the limit, new jobs wait and start as soon as a running one ends. Need more? Email [email protected].