Larger runners require a GitHub Team or Enterprise Cloud plan; navigate to Organization Settings > Actions > Runners > New runner > New GitHub-hosted runner
Choose an OS image (Ubuntu, Windows, or macOS) and a hardware size from the available vCPU/RAM/SSD options; GPU-powered sizes appear under a separate GPU tab
Set maximum concurrency (1–500 jobs) and assign the runner to an existing runner group, which controls which repositories can use it
Optionally enable static IP addresses (Enterprise Cloud only) so outbound traffic originates from a predictable IP range for firewall allowlisting
After creation, reference the runner in workflows via the auto-generated custom label shown in the runner's settings page: runs-on: ubuntu-x64-8core-32gb
Adjust runner group access under Organization Settings > Actions > Runner groups to grant or restrict specific repositories
Known gotchas
Larger runners are billed per minute at rates higher than standard GitHub-hosted runners; review the current pricing page before enabling high-concurrency limits
The 4-vCPU Windows larger runner only supports the Windows Server 2025 or Base Windows 11 Desktop image; selecting other images silently reverts the size
macOS larger runners are arm64 (Apple Silicon) by default; workflows that invoke x86-only binaries will fail without Rosetta or an x64 runner selection
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp