GitHub Actions runner specs: CPU, memory and disk for every hosted runner

What hardware does a GitHub Actions runner give you?

On a private repository, GitHub’s standard ubuntu-latest runner is a 2 vCPU x64 virtual machine with 8 GB of memory and a 14 GB SSD; on a public repository the same label gets 4 vCPU and 16 GB. Windows and Linux arm64 runners come in the same sizes, macOS on Apple M1 gets 3 cores and 7 GB, and anything bigger is a larger runner, from 4 cores and 16 GB up to 96 cores and 384 GB, for organizations on Team or Enterprise Cloud. GitHub does not name the CPU. Sengi runs 2, 4 and 8 vCPU runners on an AMD Ryzen 9 7950X3D boosting past 5 GHz, the same CPU on every job, with local NVMe: up to 3× the multi-core and 1.65× the single-core CPU of GitHub’s 2 vCPU runner, for $0.003 a minute against $0.006.

Last checked October 23, 2026, against GitHub’s hosted-runner reference, its larger-runner specifications and its Actions runner pricing reference.

GitHub-hosted runner specs, public and private repositories

GitHub sizes its standard runners by repository visibility. The same label gets a bigger machine on a public repository than on a private one:

Standard runners on public repositories are not billed. On private repositories the rate applies after the plan’s included minutes; see how much GitHub Actions minutes cost. ubuntu-slim is a container on a shared VM rather than a full VM: it has no Docker, runs unprivileged and stops any job at 15 minutes, so it suits light tasks such as labelling and linting, not builds.

GitHub larger runner specs

Past the standard sizes, GitHub sells larger runners with 4 GB of memory per core. They are open only to organizations on GitHub Team or Enterprise Cloud and are billed from the first minute:

Every size, the plan rule and the half-price alternative: GitHub larger runners: pricing and specs, and whether they pay off: are GitHub larger runners worth it?

What CPU does a GitHub Actions runner use?

GitHub publishes cores, memory and disk, not a CPU model. Its Linux and Windows runners are virtual machines on cloud server CPUs, and which one a job lands on varies: the jobs we measured ran on four different server CPU models. That is why the same job can take noticeably different times from run to run.

Sengi publishes the CPU, and it is the same on every job: an AMD Ryzen 9 7950X3D desktop CPU boosting past 5 GHz, on bare metal, with DDR5 memory and local NVMe.

All three are Ubuntu 24.04 x64, a fresh VM for every job, on personal and organization repositories alike. Measured against GitHub’s runner of the same size: up to 3× the multi-core and 1.65× the single-core CPU at 2 vCPU, 15× the random-read disk speed at 4 vCPU (296k against 19.6k IOPS), and a production Docker image built 1.9× faster at 74% less per build. More in faster GitHub Actions runners and the runner comparison.

How do I check a runner’s specs from inside a job?

Add a step that prints what the machine actually has. It works on any Linux runner:

      - name: Runner specs
        run: |
          nproc
          free -h
          df -h /
          lscpu | grep 'Model name'

nproc prints the cores the job can use, free -h the memory, df -h / the disk and what is left on it, and lscpu the CPU model the job landed on.

Which runner size do I need?

• 2 vCPU, 8 GB: lint, unit tests, small builds and the checks on every pull request.
• 4 vCPU, 16 GB: test suites that run in parallel, TypeScript and Rust builds, most Docker images.
• 8 vCPU, 32 GB: heavy Docker builds, large compiles and anything that runs out of memory at 16 GB.

More cores only help a job that uses them. A step bound to one thread gets faster from a faster core, not from more of them; see can I use a faster CPU for GitHub Actions?

Frequently asked questions

What are the specs of ubuntu-latest?

On a private repository, ubuntu-latest is a 2 vCPU x64 virtual machine with 8 GB of memory and a 14 GB SSD. On a public repository the same label gets 4 vCPU and 16 GB, with the same 14 GB SSD.

Does ubuntu-latest have 4 cores?

Only on public repositories. GitHub gives public-repository jobs 4 vCPU and 16 GB, and private-repository jobs 2 vCPU and 8 GB. For 4 cores on a private repository you need a larger runner, at $0.012 a minute on GitHub Team or Enterprise Cloud, or a third-party runner such as Sengi's sengi-standard-4-ubuntu-2404 at $0.006.

How much disk space does a GitHub Actions runner have?

GitHub lists a 14 GB SSD for every standard hosted runner, Linux, Windows and macOS alike. Larger runners come with more: 75 GB at 2 cores, 150 GB at 4, 300 GB at 8, up to 2,040 GB at 64 and 96 cores.

What CPU do GitHub Actions runners use?

GitHub does not publish a CPU model for its Linux and Windows hosted runners; they run on cloud server CPUs, and the jobs we measured landed on four different models. macOS arm64 runners use Apple M1. Sengi runs every job on the same AMD Ryzen 9 7950X3D desktop CPU, boosting past 5 GHz.

How much RAM does a GitHub Actions runner have?

8 GB on a private-repository Linux or Windows runner, 16 GB on a public-repository one, 5 GB on ubuntu-slim, 7 GB on macOS M1 and 14 GB on macOS Intel. Larger runners have 4 GB per core, from 16 GB at 4 cores to 384 GB at 96 cores.

Can I get a bigger GitHub Actions runner on a personal account?

Not from GitHub: larger runners are only for organizations on GitHub Team or Enterprise Cloud. Sengi runs 2, 4 and 8 vCPU runners on personal repositories as well as organization repositories, from $0.003 a minute.