---
title: "GitHub Actions runners in the EU, per minute | A-Line"
description: "GitHub Actions runners on A-Line.cloud from €0.002 a minute: change one line, get a fresh machine per job in the EU, pay per started minute in euro."
canonical: https://a-line.be/actions/
lang: en
---

> Index of every page on this site: https://a-line.be/llms.txt

# GitHub Actions runners from €0.002 a minute

A-Line.cloud runs your GitHub Actions jobs on a fresh machine for every job, in the EU. You change one line in your workflow, and you pay per started minute from a balance in euro.

[Request access](https://a-line.be/request-access/)

[Compare runner prices](https://a-line.be/compare/ci-runners/)

.github/workflows/ci.yml

```
jobs:
  test:
    runs-on: ubuntu-latest
    runs-on: a-line-4vcpu
    steps:
      - uses: actions/checkout@v4
      - run: pnpm install
      - run: pnpm test
```

Steps, actions and secrets stay as they are.

## One rate per vCPU, at every size

€0.001 per vCPU a minute, whatever size you pick. A bigger runner costs no more per core, so a job that runs in parallel finishes sooner for about the same money. Each size comes with 4 GB of memory per vCPU and a 50 GB disk.

| Label            | Machine          | Per minute   |
| ---------------- | ---------------- | ------------ |
| `a-line-2vcpu`   | 2 vCPU, 8 GB     | €0.002 / min |
| `a-line-4vcpu`   | 4 vCPU, 16 GB    | €0.004 / min |
| `a-line-8vcpu`   | 8 vCPU, 32 GB    | €0.008 / min |
| `a-line-16vcpu`  | 16 vCPU, 64 GB   | €0.016 / min |
| `a-line-32vcpu`  | 32 vCPU, 128 GB  | €0.032 / min |
| `a-line-64vcpu`  | 64 vCPU, 256 GB  | €0.064 / min |
| `a-line-128vcpu` | 128 vCPU, 512 GB | €0.128 / min |
| `a-line-160vcpu` | 160 vCPU, 640 GB | €0.160 / min |

A new organisation starts jobs of up to 16 vCPU. The bigger sizes open up as your limit rises with use, or sooner on request.

### Advice on the size each job needs

A runner that is too big costs money for nothing, and one that is too small keeps you waiting. While a job runs, its runner reports how much CPU, memory and disk it uses, and A-Line.cloud holds each job's recent runs against its size.

A job that keeps working at its limits, or whose last run ran out of memory, gets the advice to go one size up, so it finishes sooner. A job that leaves most of its runner idle gets the advice to go one size down, with what that saves. The advice waits in the console, and your workflow stays as it is.

ci.yml

e2e`a-line-4vcpu`

CPU

Memory

Advice: run this job on `a-line-8vcpu`

Near its limits in most of its recent runs.

An example. You change `runs-on` yourself.

**Per started minute, per job**

A 40-second job costs one minute and a 61-second job two. Minutes count per job, never for the whole workflow run.

**Waiting is free**

Time in the queue, the machine starting up and jobs that fail because of us cost nothing.

**10 GB of cache included**

actions/cache works as it is, with no change to your workflow. Past 10 GB, the cache costs €0.05 per GB a month.

**More disk when a job needs it**

Disks of 100 to 400 GB, at €0.0002 per extra GB for each started hour.

Prices excluding VAT, as of November 9, 2026. They come from the A-Line.cloud pricing API, which this page reads when it is built. [Pricing API (JSON)](https://api.a-line.cloud/v1/pricing)

## An example month

An agency with twenty client repositories runs about 150 jobs a working day: tests, builds and a deploy on every push. Each job takes 4 minutes on 4 vCPU.

1. CI jobs

   €50.40

   3,150 jobs of 4 minutes, 21 working days

   12,600 minutes at €0.004 / min

2. Cache

   Included

   Dependencies and Docker layers, under 10 GB

Your month

€50.40

Excluding VAT, from your prepaid balance.

### The same month elsewhere

The same 12,600 minutes on each provider's 4 vCPU Linux runner, at its list price and in the currency it bills in. Free minutes left out, as they differ per plan.

| Provider                                            | 4 vCPU, per minute | Example month | Source                                                                                                    |
| --------------------------------------------------- | ------------------ | ------------- | --------------------------------------------------------------------------------------------------------- |
| A-Line.cloud                                        | €0.004 / min       | €50.40        | [Source (A-Line.cloud)](https://api.a-line.cloud/v1/pricing), November 9, 2026                            |
| [Blacksmith](https://a-line.be/compare/blacksmith/) | US$0.008 / min\*   | US$100.80     | [Source (Blacksmith)](https://www.blacksmith.sh/pricing), November 10, 2026                               |
| [GitHub](https://a-line.be/compare/github-actions/) | US$0.012 / min     | US$151.20     | [Source (GitHub)](https://docs.github.com/en/billing/reference/actions-runner-pricing), November 10, 2026 |
| [Ubicloud](https://a-line.be/compare/ubicloud/)     | US$0.004 / min     | US$50.40      | [Source (Ubicloud)](https://www.ubicloud.com/docs/about/pricing), November 10, 2026                       |

Worked out from Blacksmith's listed price: Blacksmith lists one rate and counts a bigger runner as that many more minutes.

[Every size side by side](https://a-line.be/compare/ci-runners/)

## Moved over in three steps

1. ### Turn on Actions

   In the console, open Actions and choose Enable. This installs the A-Line GitHub App on your GitHub organisation.

2. ### Choose repositories

   Pick the repositories that may use the runners. Jobs from any other repository never reach us.

3. ### Change one line

   Set runs-on to a label such as a-line-4vcpu. Your next push runs on A-Line.cloud.

The CI of this website runs on a-line-4vcpu.

## What every job runs on

The details, for the developer who checks before moving over.

**A fresh machine per job**

Each job gets its own virtual machine, deleted when the job ends. Nothing carries over from one job, or one client, to the next.

**In the EU only**

Jobs run in data centres in France, Germany and Poland. A job never runs outside the EU.

**Ubuntu 24.04, Docker included**

The image follows the tools of GitHub's ubuntu-latest. Docker runs inside every job, so docker build and docker compose work as they do now.

**Secrets stay with GitHub**

GitHub hands each job its secrets, as it does on its own runners. A-Line.cloud never stores them.

**No way in**

The machine accepts no inbound connections and has no SSH. It only reaches out, to GitHub and to the cache.

**Forks stay out**

Pull requests from forks of a public repository don't run on your balance unless you turn that on.

## Questions people ask

Something else on your mind?

Write to us and Arthur replies, within 24 hours on working days.

<hello@a-line.be>

[Or see the contact page](https://a-line.be/contact/)

### Do I have to change my workflows?

Only the runs-on line of each job you move. Steps, actions and secrets stay as they are.

### Can some jobs stay on GitHub's own runners?

Yes. Only jobs with an a-line- label run on A-Line.cloud. Every other job runs where it did before, so you can move one repository at a time.

### How many jobs run at once?

A new organisation runs up to 4 jobs and 16 vCPU at once. As your account builds trust through use and payments, the limit is raised automatically: 8 jobs and 32 vCPU once you have paid €50, 16 jobs and 64 vCPU once you have paid €250, 32 jobs and 128 vCPU once you have paid €1,000 and no limit from €5,000. Need more sooner? Ask, and we raise it by hand.

### Does A-Line.cloud change a job's runner size for me?

No. When a job's recent runs show it needs a bigger or a smaller runner, the console shows the advice and the owner and admins of your organisation get a notice. You change runs-on yourself, or snooze the advice. It clears itself once it no longer holds, and it doesn't flip the other way within a month.

### What happens when my balance runs out?

Jobs only start while your balance is positive. At zero, new jobs wait in the queue and start as soon as you top up. A job that is already running finishes, and its minutes come off afterwards, so that last run can take your balance just below zero.

### Is there a free allowance?

No. You pay from the first minute, at the rate on this page, from a balance you top up in euro.

### Does it work with GitLab or Forgejo?

Not for now. The runners work with GitHub Actions only.

## How do I get in?

A-Line.cloud is invite-only while we take on new customers one at a time. Here is what happens.We reply within 24 hours on working days.

[Request access](https://a-line.be/request-access/)

Already hosted somewhere else? During early access, we do the move for you.

1. Today

   ### You tell us who you are

   A few lines about you and what you want to run.

2. Within 24 hours on working days

   ### We reply

   With questions, or with your invite.

3. After your invite

   ### You go live

   Sign in, and we help you with your first deployment.

## Key facts

- Drop-in runners for GitHub Actions through a GitHub App: a job moves when you change its `runs-on` to an `a-line-<N>vcpu` label.
- A fresh machine for every job, deleted when it ends: Ubuntu 24.04 with Docker. No inbound traffic and no SSH.
- Secrets and the workflow stay with GitHub, which hands the job what it needs.
- Linux x64 only: no arm64, Windows or macOS.
- Sizes from 2 to 160 vCPU, with 4 GB of memory per vCPU and 50 GB of disk. Extra disk from 100 to 400 GB.
- Jobs run only in the EU, in France, Germany and Poland, with no fallback outside it.
- Paid per started minute, per job, from a prepaid balance in euro. Time in the queue and start-up is free. There is no free allowance.
- A new organisation runs 4 jobs and 16 vCPU at once; the limit goes up as you pay more over time.
- Works with `actions/cache` v4 and later: a free amount per organisation, then per GB-month.
- Pull requests from forks of public repositories don't run unless you turn them on.
- Size advice per job (repository, workflow and job name, matrix jobs included): while a job runs, its runner reports CPU, memory and disk use, and A-Line.cloud looks at the job's last 7 runs. When 5 of those 7 run near the runner's CPU, memory or disk limit, or the newest run ran out of memory, it advises one size up. When CPU and memory stay mostly idle, it advises one size down and shows the estimated saving.
- The advice shows in the console, where the Actions page shows for each workflow how much of its runner every job uses; the organisation's owner and admins get a notice. Advice can be snoozed, clears itself when it no longer holds and doesn't flip direction within 30 days. It never changes a workflow: you change `runs-on` yourself.
