What is the difference between VM runners and dedicated runners?

VM runners are virtual machines in fixed sizes that we run for you on our platform. They suit pipelines with moderate demand. Dedicated runners are whole physical servers that belong to you alone: more performance and GPU options. You find the VM runner prices on the Managed GitLab page in the runner section.

Does the runner work with gitlab.com and self-managed GitLab?

Yes. It is a standard GitLab runner. You register it in your instance: on gitlab.com, in a self-managed GitLab or in a Managed GitLab that we operate.

How do GPU drivers and CUDA work?

We install and maintain the NVIDIA drivers on the host and expose the GPU to the Docker executor. You choose the CUDA version in your images. If you need a specific driver version, tell us in your request.

Where does the hardware run?

On dedicated hardware in European data centres. We name the location in our proposal. On request we operate in German data centres only.

Managed CI capacity

Dedicated GitLab Runners: your CI on dedicated hardware, run by us

The whole machine is your team's: no noisy neighbours, no per-minute bill, predictable build times. We provision the runner host, patch it, monitor it and replace it when needed.

You register the runner in your GitLab: hosted by us, on gitlab.com or self-managed. Optionally it joins your network over VPN.

Submit a request Classes

Dedicated, CPU and GPU
Hardware: EU (Germany only on request)
Data location: Monthly, no build minutes
Billing

WHY DEDICATED

CI capacity you can count on

Predictable build times: The runner shares compute with nobody. A build takes as long today as it did yesterday.
No minute meter: You pay a fixed monthly price. Long pipelines and many parallel jobs cost no extra.
Data in the EU: Your dedicated hardware runs in European data centres. We name the location in our proposal. On request in German data centres only.
Inside your network via VPN: Optionally connected over WireGuard or IPsec: builds reach internal registries, artifact stores and test systems without you opening anything to the internet.

CLASSES AND PRICES

Dedicated runners in fixed sizes

Each class is a server of its own for your pipelines. Prices are per month.

All prices net, excl. VAT. Monthly term. Term discounts on request.

CPU runners
Class	Specification	Suited for	Price
Dedicated 8	8 cores, 64 GB RAM, 2× NVMe	Small teams, steady pipelines	from €249 / month
Dedicated 16	16 cores, 128 GB RAM, NVMe	Most teams: Docker builds, test suites	from €649 / month
Dedicated 48	48 cores, 128 to 512 GB RAM, NVMe	Monorepos, heavy compiles, many parallel jobs	from €1,249 / month
Dedicated Custom	On request: RAM, disks, 10G, location	Special requirements	Submit a request

The GPU options name the card directly. The price covers the whole server with one card.

GPU runners
Class	Specification	Suited for	Price
GPU RTX PRO 4000	NVIDIA RTX PRO 4000 Blackwell, 24 GB VRAM	CUDA builds, ML tests, small inference tests	from €549 / month
GPU L40S	NVIDIA L40S, 48 GB VRAM	Inference and fine-tuning tests, rendering	from €1,190 / month
GPU RTX PRO 6000	NVIDIA RTX PRO 6000 Blackwell, 96 GB VRAM	Large models in CI, fine-tuning	from €1,990 / month
GPU H100	NVIDIA H100 (NVL), 94 GB VRAM	Training jobs, heavy ML pipelines	from €2,090 / month
GPU H200	NVIDIA H200 (NVL), 141 GB VRAM	Largest models, long training runs	from €3,890 / month
Multi-GPU	8× H100, H200 or RTX PRO 6000 (NVLink)	Distributed training, large pipelines	Contact sales

Contact sales

WHAT'S INCLUDED

Included in every class

Provisioning and registration of the GitLab runner (Docker executor, Kubernetes executor on request)
OS and runner updates, security patches, monitoring and alerting; hardware swap on failure
Build cache on local NVMe, optional S3 cache on our object storage
Optional VPN connection into your network (VPN prices per the price list)
Support per your existing package or support plan
Monthly term, term discounts on request

Not included: Build-minute billing (there is none), GitLab licences and customer-specific tooling in images (on request at the hourly rate).

HOW IT WORKS

Your own runner in four steps

01

Request

You tell us about your pipelines, your GitLab and your wishes for location and VPN.
02

Proposal

We recommend a class and a location and give you the price.
03

Provisioning

We set up the runner and register it. Lead time depends on the hardware supplier, typically a few days.
04

Use it monthly

From the start your runner runs on a monthly term. We handle operations and maintenance.

FAQs

Short answers about dedicated runners.

See the VM runner prices

: VM runners are virtual machines in fixed sizes that we run for you on our platform. They suit pipelines with moderate demand. Dedicated runners are whole physical servers that belong to you alone: more performance and GPU options. You find the VM runner prices on the Managed GitLab page in the runner section.
: Yes. It is a standard GitLab runner. You register it in your instance: on gitlab.com, in a self-managed GitLab or in a Managed GitLab that we operate.
: We install and maintain the NVIDIA drivers on the host and expose the GPU to the Docker executor. You choose the CUDA version in your images. If you need a specific driver version, tell us in your request.
: On dedicated hardware in European data centres. We name the location in our proposal. On request we operate in German data centres only.

Do you have questions or would you like a personalized offer? We are happy to advise you.

Submit Inquiry

Contact

Our cloud experts are happy to provide personalized advice.

Sillemstraße 76A

20257 Hamburg, Deutschland

Mon - Fri: 09:00 AM - 06:00 PM

+49 40 239 69 754 0

hello@bnerd.com