SKIP_TO_MAIN_CONTENT
02.0 // CPU interface x86_64 · arm64 From 2 to 64 cores

Faster CI.
More cores.
Less waiting.

On-demand x64 and ARM64 runners from 2 to 64 vCPU. Spot pricing — up to 85% cheaper than GitHub hosted. If you were on BuildJet, you can migrate in about five minutes.

X64_SPOT_2V
$0.00032 /min
ARM64_SPOT_2V
$0.00026 /min
MAX_VCPU
64
COLD_START
<60 s
01 Who this is for
[REF_010]

Bottlenecked
by compute?

If your CI is bottlenecked by compute, more cores fix it.

[01] >_

Large parallel test suites that take 20+ minutes on a 2-core runner.

[02] >_

Docker image builds that are CPU-bound and benefit from more cores.

[03] >_

Compilation-heavy builds (Rust, C++, Go) that scale with core count.

[04] >_

ARM-native testing for software that ships to Graviton or Apple Silicon.

> FOOTNOTE_001

The typical alternative is either waiting on slow runners or provisioning your own and absorbing the maintenance overhead. Neither is great.

02 Live pricing
[REF_PRICING] // LIVE_DATA

CPU Pricing

Live per-minute rates for x64 and ARM64 CPU runners. Auto-refreshes every 30 seconds.

03 Yaml examples
[REF_030]

x64
or ARM64.
Pick one.

Add an architecture=arm64 flag and you get Graviton instead of Intel/AMD. That's the entire difference.

EXAMPLE_01.YAML // X64_PARALLEL_TESTS 8_VCPU
jobs:
test:
runs-on: [machine, cpu=8]
steps:
- uses: actions/checkout@v4
- uses: actions/setup-node@v4
with:
node-version: '20'
- run: npm ci
- run: npm test -- --maxWorkers=8
EXAMPLE_02.YAML // ARM64_NATIVE_BUILD GRAVITON
jobs:
build-arm:
runs-on: [machine, cpu=8, architecture=arm64]
steps:
- uses: actions/checkout@v4
- run: cargo build --release
- uses: actions/upload-artifact@v4
with:
name: binary-arm64
path: target/release/myapp
04 Use cases
[REF_040] PARALLEL_TESTS

Large parallel test suites

If your test suite takes 20 minutes on a 2-core runner, a 16-core machine with more RAM will cut that down significantly. Most test frameworks can parallelise across cores with a single flag.

[REF_041] DOCKER_BUILDS

Fast Docker builds

Docker layer building is CPU-bound and benefits directly from more cores. A 32-core runner can cut a large image build from 15 minutes to 3.

[REF_042] COMPILATION

Compilation-heavy builds

Rust, C++, Go — anything with heavy compile steps scales almost linearly with core count up to a point.

[REF_043] ARM_NATIVE

ARM-native testing

If you're shipping software that runs on ARM (Apple Silicon, AWS Graviton, Raspberry Pi), testing on a real ARM runner catches issues that cross-compilation misses.

05 Initialize
READY
> CPU_READY // AWAITING_INPUT

More cores.
Fewer minutes.

$10 free compute. Connect your GitHub org in two minutes. $10 free compute, no credit card required.