Skip to main content

Larger runners reference

Find information about larger runners, including their specifications and customization options.

Machine sizes for larger runners

You can choose from several specifications for larger runners.

Specifications for general larger runners

CPUMemory (RAM)Storage (SSD)ArchitectureOperating system (OS)
514 GB14 GBarm64 (M2)macOS
1230 GB14 GBx64 (Intel)macOS
28 GB75 GBx64, arm64Ubuntu
416 GB150 GBx64, arm64Ubuntu, Windows
832 GB300 GBx64, arm64Ubuntu, Windows
1664 GB600 GBx64, arm64Ubuntu, Windows
32128 GB1200 GBx64, arm64Ubuntu, Windows
64208 GB2040 GBarm64Ubuntu, Windows
64256 GB2040 GBx64Ubuntu, Windows
96384 GB2040 GBx64Ubuntu, Windows

Note

The 4-vCPU Windows runner only works with the Windows Server 2025 or the Base Windows 11 Desktop image.

Note

The 5-vCPU macOS runner is in public preview and subject to change.

Specifications for GPU larger runners

CPUGPUGPU cardMemory (RAM)GPU memory (VRAM)Storage (SSD)Operating system (OS)
41Tesla T428 GB16 GB176 GBUbuntu, Windows

Runner images

Larger runners run on virtual machines (VMs), and GitHub installs a virtual hard disk (VHD) on this machine during the VM creation process. You can choose from different VM images to install on your runners.

GitHub-owned images: These images are maintained by GitHub and are available for Linux x64, Windows x64, and macOS (x64 and arm) runners. For more information on these images and a full list of included tools for each runner operating system, see the GitHub Actions Runner Images repository.

Partner Images: Partner images are not managed by GitHub and are pulled from the Azure Marketplace. See below for resources on where to find more information and to report issues for partner images.

Available macOS larger runners and labels

The following machines are available for macOS larger runners.

Runner SizeArchitectureProcessor (CPU)Memory (RAM)Storage (SSD)Workflow label
LargeIntel1230 GB14 GBmacos-latest-large, macos-13-large, macos-14-large [latest], macos-15-large
XLargearm64 (M2)5 (+ 8 GPU hardware acceleration)14 GB14 GBmacos-latest-xlarge, macos-13-xlarge, macos-14-xlarge [latest], macos-15-xlarge

Note

The XLarge macOS runner is in public preview and subject to change.

Limitations for macOS larger runners

  • All actions provided by GitHub are compatible with arm64 GitHub-hosted runners. However, community actions may not be compatible with arm64 and need to be manually installed at runtime.
  • Nested-virtualization and Metal Performance Shaders (MPS) are not supported due to the limitation of Apple's Virtualization Framework.
  • Networking capabilities such as Azure private networking and assigning static IPs are not currently available for macOS larger runners.
  • The arm64 macOS runners do not have a static UUID/UDID assigned to them because Apple does not support this feature. However, Intel MacOS runners are assigned a static UDID, specifically 4203018E-580F-C1B5-9525-B745CECA79EB. If you are building and signing on the same host you plan to test the build on, you can sign with a development provisioning profile. If you do require a static UDID, you can use Intel runners and add their UDID to your Apple Developer account.

Networking for larger runners

By default, larger runners receive a dynamic IP address that changes for each job run. Optionally, GitHub Enterprise Cloud customers can configure their larger runners to receive static IP addresses from GitHub's IP address pool. For more information, see About GitHub's IP addresses.

When enabled, instances of the larger runner will receive IP addresses from specific ranges that are unique to the runner, allowing you to use the ranges to configure a firewall allowlist. You can use up to 10 larger runners with static IP address ranges in total across all your larger runners. For more information, see Managing larger runners.

If you would like to use more than 10 larger runners with static IP address ranges, please contact us through the GitHub Support portal.

Note

If runners are unused for more than 30 days, their IP address ranges are automatically removed and cannot be recovered.