Faster GitHub Actions runners: what makes a runner fast, measured

How do I get faster GitHub Actions runners?

Run your jobs on faster hardware by changing one runs-on line. Sengi runs every GitHub Actions job on an AMD Ryzen 9 7950X3D desktop CPU boosting past 5 GHz, on bare metal with local NVMe, in a fresh VM for every job, and 2 vCPU jobs start on a pre-booted VM in 2 seconds. Against GitHub’s runner of the same size, that is up to 3× the multi-core and 1.65× the single-core CPU, 15× the random-read disk speed, and builds up to 1.9× faster. It costs less, not more: $0.003 a minute against GitHub’s $0.006, up to 74% less per build.

Last checked October 23, 2026. Speed figures measured on Sengi October 6 and 9, 2026, against GitHub’s runners of the same size; GitHub’s specifications and rates from its hosted-runner and runner pricing references, checked October 23, 2026.

What makes a GitHub Actions runner fast?

Four things decide how long a job spends on the machine: how fast each CPU core is, how many cores the build can use, how fast the disk reads and writes, and how long the job waits for a machine to start. Here is each one on GitHub’s standard runner and on Sengi.

How much faster are Sengi runners, measured?

On RunsOn’s open benchmark harness, run on Sengi October 9, 2026 at 4 vCPU, against RunsOn’s published results for GitHub’s 4 vCPU runner:

The Docker build: 1.9× faster at 74% less per build. On the same 2 vCPU as GitHub’s ubuntu-latest, real CI jobs ran up to 1.8× faster: a Node and Docker build took 299 seconds on Sengi against 539 on GitHub, measured October 6, 2026, and cost $0.015 a run against $0.054. The step-by-step results for five open-source projects are in how to speed up GitHub Actions.

Do faster GitHub Actions runners cost more?

Not on Sengi. Faster hardware finishes the job in fewer minutes, and each minute costs less:

Up to half GitHub’s price per minute, and up to 74% less per build once the shorter run time is counted. 3,000 minutes are included on Sengi every month. GitHub’s 4 and 8 vCPU runners are larger runners, open only to organizations on Team or Enterprise Cloud; Sengi runs every size on personal repositories too. See Sengi runner pricing and the runner comparison.

Which jobs get faster on a faster runner?

• Compiles and type checks. Rust, Go, C++, TypeScript and Java builds use every core and reward single-core speed: the Rust build ran 1.6× faster and the TypeScript typecheck 1.5× faster.
• Docker image builds. CPU, disk and layer cache together: 1.9× faster on the runner, then 1.9× faster again warm with Docker layer caching.
• Dependency installs and test suites. npm ci, bundle install and test databases are disk-bound, and local NVMe reads 15× faster at random.
• The pull requests your coding agents open all day. Every one waits on CI; a 2-second start and a faster build shorten each loop.

Waiting in a queue, waiting on an external service or a chain of serial jobs is a workflow fix: diagnose stuck jobs and shorten the critical path, then run what is left on a faster machine.

How do I switch to a faster runner?

Change runs-on and merge. Your workflows stay GitHub Actions workflows.

jobs:
  build:
-   runs-on: ubuntu-latest
+   runs-on: sengi-standard-2-ubuntu-2404   # or sengi-standard-4-ubuntu-2404, sengi-standard-8-ubuntu-2404

Or give your coding agent one prompt: it finds every job Sengi can run, changes the label and opens the pull request. See how switching works. Coming from another runner service? See Blacksmith, Depot, Namespace, WarpBuild or RunsOn.

Frequently asked questions

How do I get faster GitHub Actions runners?

Change the machine behind runs-on. Sengi runs every job on an AMD Ryzen 9 7950X3D desktop CPU boosting past 5 GHz, on bare metal with local NVMe, in a fresh VM, with 2 vCPU jobs ready in 2 seconds. Set runs-on to sengi-standard-2-ubuntu-2404, sengi-standard-4-ubuntu-2404 or sengi-standard-8-ubuntu-2404 and merge; builds finish up to 1.9× faster than on GitHub's runners.

Why is GitHub Actions so slow?

On a private repository, ubuntu-latest is a 2 vCPU, 8 GB virtual machine on a cloud server CPU, and every job starts with an empty disk and a cold Docker engine. CPU-heavy builds, disk-heavy installs and Docker builds without a layer cache all wait on that hardware. Queue time, serial job dependencies and repeated setup add to it, and the speed-up guide covers those.

Are Sengi runners faster than GitHub's?

Yes. On RunsOn's open benchmark harness, run October 9, 2026 at 4 vCPU, Sengi built a production Docker image 1.9× faster than GitHub's runner and ran Rust, TypeScript and Docker builds together 1.8× faster, with 15× the random-read disk speed. On the same 2 vCPU as ubuntu-latest, it delivers up to 3× the multi-core and 1.65× the single-core CPU.

Do faster runners cost more?

Not on Sengi. It charges $0.003 a minute for 2 vCPU against GitHub's $0.006, $0.006 for 4 vCPU against $0.012, and $0.012 for 8 vCPU against $0.022. A faster job bills fewer minutes too, so each build costs up to 74% less than on GitHub's runner.

Will a faster runner speed up Docker builds?

Yes, twice over. Sengi's CPU and local NVMe built a production Docker image 1.9× faster than GitHub's 4 vCPU runner, and Docker layer caching through type=gha is included, which took a warm Docker build from 68 seconds to 36 seconds on Sengi.

Do I have to change my workflows?

Only the runs-on line. Your workflows, required checks, secrets and logs stay in GitHub Actions. Your coding agent can make the change and open the pull request from the Sengi prompt.