> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trunk.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Trunk Runners

> Fast, reliable, affordable runners for GitHub Actions.

Trunk Runners give each GitHub Actions job its own VM, a fast SSD, and a cache
near the runner. Your workflows stay on GitHub. You change one line, the
`runs-on` label, and Trunk starts a fresh VM for the job and removes it when the
job ends.

> [!NOTE]
>
> Trunk Runners is in private beta. Contact
> [support@trunk.io](mailto:support@trunk.io) to join.

> [card] Fast
>
> 2x faster hardware, 4x faster caches, and templates that skip setup.

> [card] Reliable
>
> A managed fleet, with priority for the branches that must stay fast.

> [card] [Affordable](https://docs.trunk.io/runners/pricing.md)
>
> Pay per CPU and GB minute. Size each job to what it needs.

## Fast

* **Faster hardware.** Jobs run on bare metal with a local SSD: 2x faster than
  GitHub-hosted runners, and 60% faster than self-hosted runners on AWS.
* **Templates.** Bake your setup steps into a [template](https://docs.trunk.io/runners/actions/templates.md)
  so a job starts ready, instead of installing the same tools on every run.
* **In-cluster cache.** `actions/cache`, Bazel, sccache, and Nix use a shared
  cache in the same cluster as your jobs, with no workflow changes. Downloads
  are 4x faster than the GitHub Actions cache. See
  [Caching](https://docs.trunk.io/runners/cache/index.md).

## Reliable

* **No infrastructure to operate.** Trunk runs the fleet, images, and caches.
  No self-hosted runners or controller to maintain.
* **Priority branches.** Merge queue, default, and release branches start
  first when CI is busy, so the queue does not wait behind pull request jobs. See
  [Quota and runner priority](https://docs.trunk.io/runners/settings/limits-and-priority.md).
* **Nothing dropped.** At your CPU quota, new jobs wait and start as CPUs free
  up.

## Affordable

* **Usage-based.** You pay for the CPUs and memory a job's machine has while
  the job runs. The default machine costs \$0.006 per minute. See
  [Pricing](https://docs.trunk.io/runners/pricing.md).
* **Right-sized.** Mix CPU and memory per job, so a lint job does not pay for a
  build machine. See [Machine sizes](https://docs.trunk.io/runners/actions/sizes.md).
* **Run less CI.** Pair with [Dynamic CI](https://docs.trunk.io/ci/dynamic-ci.md) to skip jobs a
  change cannot affect.

> [card] [Get started](https://docs.trunk.io/runners/getting-started.md)
>
> Turn on Trunk Runners for your organization and run your first job.
