GitHub Actions runners with the best single thread CPU
Avrea Linux arm64 leads the measured results at 52,794 iterations/s. 26 of 26 configurations have a score. This ranks the imported tests, not full workflow runtime.
A serial build step can keep an entire workflow waiting even when the machine has spare cores. This ranking compares how quickly one execution thread completes the same CoreMark workload on the runners in our independent benchmark. It includes all tested operating systems and binary architectures, with the highest measured median first. The leaders below come directly from the current imported results.
By TechCompare · Updated
Current leaders
Results imported 2026-10-10. Pricing checked 2026-10-10. Higher is better. All operating systems and architectures are included. Prices and value use USD. RunJob's EUR rate is converted at €1 = $1.1206 (ECB, 2026-10-09).
Calculator
Ranked runner results
| RunnerRunner | CPUCPU scoreSingle thread | Cost/vCPU/minCost/vCPU/minUSD | Max vCPUMax runner vCPU |
|---|---|---|---|
| Avrea Linux arm64 Apple M5 Max | 52,794 | 0.006 | 16 |
| Avrea macOS arm64 Apple M5 Max | 52,014 | 0.01 | 16 |
| Namespace macOS arm64 Apple M4 Pro | 46,338 | 0.015 | 16 |
| Blacksmith Linux x64 AMD EPYC (model hidden) | 45,851 | 0.002 | 32 |
| Avrea Linux x64 AMD EPYC 4585PX 16-Core | 43,680 | 0.002 | 32 |
| BuildPulse Linux x64 AMD EPYC 9R45 | 43,146 | 0.002 | 64 |
| Blacksmith macOS arm64 Apple M4 Pro | 41,001 | 0.0133 | 12 |
| Blacksmith Windows x64 AMD EPYC 4565P 16-Core | 40,527 | 0.004 | 32 |
| StarSling Linux x64 AMD EPYC (model hidden) | 38,218 | 0.002 | 64 |
| Ubicloud Linux x64 AMD Ryzen 9 7950X3D 16-Core | 33,393 | 0.001 | 30 |
| Namespace Linux x64 AMD EPYC (model hidden) | 33,130 | 0.0015 | 64 |
| BuildPulse Linux arm64 AWS Graviton (gen undisclosed) | 33,116 | 0.0013 | 64 |
| GitHub macOS arm64 Apple M2 Pro | 32,583 | 0.0207 | 5 |
| GitHub Linux arm64 Microsoft Cobalt 100 | 31,687 | 0.0025 | 64 |
| Avrea Windows x64 AMD EPYC 4585PX 16-Core | 31,647 | 0.004 | 16 |
| GitHub Linux x64 AMD EPYC 9V74 80-Core | 31,133 | 0.003 | 96 |
| Latchkey Linux x64 AMD EPYC 7R13 | 30,065 | 0.0013 | 16 |
| Blacksmith Linux arm64 ARM64 (model undisclosed) | 27,191 | 0.0013 | 32 |
| GitHub Linux x86 AMD EPYC 7763 64-Core | 25,491 | 0.003 | 96 |
| Namespace Linux arm64 AmpereOne family (SKU undisclosed) | 23,534 | 0.0015 | 32 |
| GitHub Windows arm64 Cobalt 100 | 22,553 | 0.005 | 64 |
| Ubicloud Linux arm64 Ampere Altra Q80-30 | 21,630 | 0.0006 | 30 |
| GitHub Windows x64 AMD EPYC 7763 64-Core | 18,327 | 0.005 | 96 |
| RunJob Linux x64 AMD EPYC 7502 32-Core | 17,973 | 0.0013 | 48 |
| GitHub Windows x86 AMD EPYC 7763 64-Core | 15,003 | 0.005 | 96 |
| GitHub macOS x64 Intel Core i7-8700B | 8,375 | 0.0155 | 12 |
Single thread CPU: CoreMark iterations/s. Disk and cache: composite MiB/s. Internet: composite Mbps. Value: score ÷ USD cost/vCPU/min. K means thousand, M means million. Higher is better. Cache uses each runner's best tested cache or artifact backend. Limits: maximum runner vCPU. Prices and value use USD. RunJob's EUR rate is converted at €1 = $1.1206 (ECB, 2026-10-09).
Open the full GitHub Actions runner benchmark toolHow the CPU ranking is calculated
The CPU column is the median CoreMark iterations per second across five accepted samples. Higher is better. Each runner uses one execution thread, regardless of its requested vCPU count. The ranking does not multiply by cores, blend in disk performance, or rescale the score against the fastest machine. Adding a new result can change a runner's position without changing its measured score. The table and the leader summary use that same calculation.
Use the CPU model as context
The runner column shows the researched CPU model beneath the provider and platform. Some guests expose a full model string, while others reveal only a family or architecture. A provider mapping is kept distinct from a model reported by the guest. Open the full benchmark tool to inspect that evidence and the exact workflow tag. A similar score is not proof that two providers use the same processor.
When another ranking is more useful
Parallel compilation can depend on the available vCPUs and memory, while dependency installation can spend much of its time waiting for disk, cache, or internet traffic. Those stages have their own ranking pages. CPU value adds the quoted rate to this single thread measurement. None of these pages measures queue time or a complete repository build, so use them to choose candidates for a workflow trial.
Measurement context
CPU speed depends on more than the processor name. The guest operating system, compiler, optimization flags, and CPU affinity all affect this test. Linux and Windows pin the execution thread to one logical CPU. macOS uses one unpinned thread that can migrate. The table includes those results together, while the full tool exposes the sample range, variation, and compiler for each configuration.
Verdict
Start with a runner near the top if serial CPU work is your main delay, then time your own build on the required platform. A high CoreMark score is useful evidence for one kind of CPU work. It does not establish the shortest complete workflow or the lowest bill.
More runner benchmark scenarios
Frequently asked questions
Which runner currently leads for single thread CPU performance?
Do these runner rankings update automatically?
Are all operating systems and architectures included?
Does a missing result mean a runner scored zero?
Related tools
Data Transfer Calculator
Estimate transfer times for files over USB, WiFi, Ethernet, and more.
Use tool ➜Memory and Storage Throughput Visualizer
Visualize the massive speed difference between CPU cache, RAM, and storage.
Use tool ➜RAM Latency Calculator
Convert DDR3/DDR4/DDR5 timings (CL, tRCD, tRP, tRAS) into true latency in nanoseconds.
Use tool ➜