CI runners,faster builds at less than half the cost
BuildPulse Runners are a drop-in replacement for GitHub Actions, Semaphore, and CircleCI hosted runners. Modern hardware, smart caching, and per-minute billing: no workflow rewrites, no self-hosting overhead. Every speed claim on this page is measured in public, reproducible benchmarks.
# .github/workflows/ci.yml jobs: test: runs-on: bp-ubuntu-latest-x64-4x # ← swap steps: - uses: actions/checkout@v4 - run: npm ci && npm test
# .semaphore/semaphore.yml agent: machine: type: bp-ubuntu-latest-x64-4x # ← swap blocks: - name: Test task: jobs: - name: npm test commands: - npm ci && npm test
# .circleci/config.yml jobs: test: machine: image: ubuntu-2404:current resource_class: buildpulse/bp-ubuntu-latest-x64-4x # ← add steps: - checkout - run: npm ci && npm test
Built for fast-moving teams
Drop-in replacement
Swap one line in your workflow config: GitHub Actions, Semaphore, or CircleCI. Your pipelines keep working, just faster.
Faster CPUs, more cores
Newer-generation AMD EPYC processors, 2.7x the measured sysbench throughput of GitHub-hosted runners. Real compile, test, and docker workloads finish 1.2-2x faster.
Half the total cost
Per-minute rates 20 to 50% below GitHub, plus measurably faster hardware, so the same workloads cost 54 to 65% less. No surprise bills, no minimums.
Smart caching
Built-in Docker layer, dependency, and build cache that survives across runs.
No self-hosting headaches
No EC2 instances to patch, no autoscaler to tune, no security boundary to worry about.
Per-org usage analytics
See exactly which workflows burn minutes and where to optimize, on every plan. GitHub Actions, Semaphore, and CircleCI minutes tracked separately.
Per-minute pricing, no minimums
Usage-based billing, no commitments. Per-minute rates run 20–50% below GitHub's larger runners, and with ~2× faster hardware, most teams land near half the total cost. arm64 is priced below x86, and Docker builders carry no premium over standard.
Standard runners: x86
| Runner | vCPU | RAM | Storage | BuildPulse $/min | GitHub Actions $/min | Savings |
|---|---|---|---|---|---|---|
| 2x | 2 | 8 GB | 75 GB | $0.004 | $0.006 | 33% |
| 4x | 4 | 16 GB | 150 GB | $0.008 | $0.012 | 33% |
| 8x | 8 | 32 GB | 300 GB | $0.016 | $0.022 | 27% |
| 16x | 16 | 64 GB | 600 GB | $0.032 | $0.042 | 24% |
| 32x | 32 | 128 GB | 1200 GB | $0.064 | $0.082 | 22% |
| 64x | 64 | 256 GB | 2400 GB | $0.128 | $0.162 | 21% |
Standard runners: arm64 (Graviton)
| Runner | vCPU | RAM | Storage | BuildPulse $/min | GitHub Actions $/min | Savings |
|---|---|---|---|---|---|---|
| 2x | 2 | 8 GB | 75 GB | $0.0025 | $0.005 | 50% |
| 4x | 4 | 16 GB | 150 GB | $0.005 | $0.008 | 38% |
| 8x | 8 | 32 GB | 300 GB | $0.010 | $0.014 | 29% |
| 16x | 16 | 64 GB | 600 GB | $0.020 | $0.026 | 23% |
| 32x | 32 | 128 GB | 1200 GB | $0.040 | $0.050 | 20% |
| 64x | 64 | 256 GB | 2400 GB | $0.080 | $0.098 | 18% |
Docker builder runners: NVMe-backed (same rate as standard)
| Runner | vCPU | RAM | Storage | BuildPulse $/min | GitHub Actions $/min | Savings |
|---|---|---|---|---|---|---|
| 4x | 4 | 16 GB | 150 GB | $0.008 | $0.012 | 33% |
| 8x | 8 | 32 GB | 300 GB | $0.016 | $0.022 | 27% |
| 16x | 16 | 64 GB | 600 GB | $0.032 | $0.042 | 24% |
| 32x | 32 | 128 GB | 1200 GB | $0.064 | $0.082 | 22% |
Visibility you do not get from your CI provider
Track minutes per workflow, per repo, and per team. See which jobs are slowest and where the cache is missing. Drill into any run to see the exact bottleneck, across GitHub Actions, Semaphore, and CircleCI.
Common questions
Do I have to change my workflows?
One line. On GitHub Actions you change the runs-on label to a BuildPulse label such as bp-ubuntu-latest-x64-4x; on Semaphore it is the machine type, and on CircleCI the resource_class. Your steps, actions, and scripts stay exactly as they are.
How are runners billed?
Per minute, with no minimums and no commitments. Per-minute rates run 20 to 50% below GitHub Actions larger runners, depending on size, and the full rate card for x86, arm64, and Docker builder runners is published on this page.
How much faster and cheaper are they really?
The hardware is newer-generation AMD EPYC with 2.7x the measured sysbench throughput of GitHub-hosted runners. Real compile, test, and docker workloads finish 1.2 to 2x faster, which combined with the lower per-minute rate lands the same workloads at 54 to 65% less total cost. The speed claims come from public, reproducible benchmarks.
What sizes are available?
Six sizes from 2 vCPU / 8 GB up to 64 vCPU / 256 GB, on x86 and arm64 (Graviton), plus NVMe-backed Docker builder runners from 4x to 32x. arm64 is priced below x86 and Docker builders carry no premium over standard runners of the same size.
Is this self-hosting?
No. There are no EC2 instances to patch, no autoscaler to tune, and no security boundary to manage on your side. BuildPulse runs the fleet, and you get per-org usage analytics showing which workflows and jobs burn minutes across GitHub Actions, Semaphore, and CircleCI.