# Tuist Runners for GitLab CI: macOS and Linux

Managed macOS and Linux runners for GitLab CI on GitLab.com and self-managed GitLab. Pick the machine with a job tag. Invite-only.

Your pipelines stay in GitLab, and Tuist provides an isolated machine for each tagged job, next to the same build cache your developers and coding agents use. Runners are invite-only while capacity scales; [request an invitation](mailto:contact@tuist.dev).

## Choose the machine with a tag

Create a project or group runner in GitLab with the Tuist profile labels as its tags, connect it once in Tuist with its authentication token, and tag each job:

```yaml
test:
  tags: [tuist-macos]
  script:
    - tuist test
  retry:
    max: 2
    when: runner_system_failure
```

No GitLab Runner to operate. Works with GitLab.com and self-managed GitLab 19.3+ over public HTTPS; see the [GitLab CI guide](https://tuist.dev/en/docs-markdown/guides/features/runners/ci-providers/gitlab-ci).

## How it differs from GitLab-hosted runners

- **No fleet to operate.** Tuist provides the macOS and Linux machines, images, and Xcode versions.
- **A build cache shared beyond CI.** Runners share the [Tuist cache](https://tuist.dev/marketing-markdown/cache) with developer machines and coding agents.
- **Directories that survive between jobs.** [Cache volumes](https://tuist.dev/en/docs-markdown/guides/features/runners/cache-volumes) keep build caches between jobs; forks never save to them.

## Sources and review

Sources checked on **2026-11-13**. Written by Tuist from its own documentation and the provider's public documentation; not an independent benchmark. Primary sources: [GitLab-hosted runners](https://docs.gitlab.com/ci/runners/hosted_runners/) and the Tuist [GitLab CI setup guide](https://tuist.dev/en/docs-markdown/guides/features/runners/ci-providers/gitlab-ci).

## Further reading

- [Tuist Runners](https://tuist.dev/marketing-markdown/runners) for all supported CI providers, machine profiles, and how runners fit with the rest of Tuist.
- Tuist Runners for [GitHub Actions](https://tuist.dev/marketing-markdown/runners/github-actions) and [Buildkite](https://tuist.dev/marketing-markdown/runners/buildkite).
- [Rising CI costs](https://tuist.dev/marketing-markdown/solutions/ci-costs) on measuring total cost before and after moving jobs.

## Frequently asked questions

### Do we have to leave GitLab CI?

No. Pipelines, variables, artifacts, and logs stay in GitLab. Tuist only provides the machine for jobs tagged with a Tuist profile.

### Does it work with self-managed GitLab?

Yes, for GitLab 19.3 or newer reachable over public HTTPS. [GitLab-hosted runners](https://docs.gitlab.com/ci/runners/hosted_runners/) cover GitLab.com and GitLab Dedicated, with macOS still in beta, so self-managed teams often run their own fleet; Tuist Runners serve both GitLab.com and reachable self-managed instances.

### What happens when we reach our concurrency limit?

Jobs Tuist has already acquired wait for a machine for up to their job timeout, at most 12 hours, and further jobs stay pending in GitLab. A job that never gets a machine fails with `runner_system_failure`, which a retry policy can handle.

### Where are the runner tokens kept?

The reusable runner authentication token stays encrypted on the Tuist server. Machines receive credentials for their assigned job only.

### Can jobs use `image:` and `services:`?

No. Jobs use GitLab's shell executor inside the isolated machine, so `image:` and `services:` do not configure the environment; use the tools on the runner image or install them in your script.

### How is it priced?

Runners are billed per minute by machine shape. See the [pricing page](https://tuist.dev/pricing) for current rates.
