Four sizes.
One platform.
Nothing is held back for the expensive plan — the difference is the size of the machine, not the list of features. Billing is hourly: stop the server and billing stops with it.
- 2 vCPU AMD Ryzen
- 4 GB RAM
- 50 GB NVMe
- Snapshots on the node: 1
- Network copies — €3 / mo extra
- Everything in the list below
- 4 vCPU AMD Ryzen
- 8 GB RAM
- 80 GB NVMe
- Snapshots on the node: 1
- Network copies — €4 / mo extra
- Everything in the list below
- 6 vCPU AMD Ryzen
- 12 GB RAM
- 100 GB NVMe
- Snapshots on the node: 2
- Network copies — €5 / mo extra
- Everything in the list below
- 8 vCPU AMD Ryzen
- 16 GB RAM
- 120 GB NVMe
- Snapshots on the node: 2
- Network copies — €6 / mo extra
- Everything in the list below
// A bigger machine than these — write to us, we will quote it
Not a handover — a project you keep
A project is never finished. Features keep being added, so a developer is always working on it — and two different people look at the same server. One watches the machine: firewall, stack, copies, whether it is still answering. The other deploys, looks at the log and rolls back when a release goes wrong.
The price is for both of those jobs, for as long as the project lives. That is why the plans are not a one-off installation fee and why the hundredth deploy is as cheap as the first.
For whoever watches the machine
Firewall from the panel, the stack installed to order, copies on a schedule, a snapshot before a risky change, and an email when the disk is filling up or the site stops answering. The role is called maintainer.
For whoever writes the code
A repository, a deploy on every push to the tracked branch, a build command of your own, the web server and system logs without SSH, and a rollback to any earlier deploy. The role is called developer.
The same on every plan
There is no cheaper plan with the tools switched off. Only the machine is smaller.
Repository and deploys
A repository on the built-in git server, access for the project team, a webhook already wired. Push to the tracked branch and the new release is built alongside the running one and switched on at the end — in one rename, so a visitor never catches a half-updated site.
Staging next to production
Two branches of one repository on one machine, each with its own directory and address. Check the layout before the client does — without paying for a second server.
Rollback to any deploy
Every deploy is recorded with its commit, and the built release is kept. Pick an earlier one and the site is back on it in a moment — nothing is fetched and nothing is rebuilt. Your repository history stays untouched.
A team with roles
Invite people by email and give each of them a role: maintainer, developer or viewer. Remove someone and access disappears everywhere at once — no forgotten keys left on machines.
Address and certificate
The panel writes the web server config, points the domain through our own DNS and issues the certificate. Until you buy a domain, the site opens on the server IP.
Firewall from the panel
Open and close ports without touching iptables. By default only what a web server and SSH need is open.
It keeps an eye on things
Disk, memory, whether the machine responds and whether the site opens from outside — checked every five minutes. One email per problem, and it closes itself when the cause is gone.
Logs without SSH
Web server errors, the system journal and your application log — readable from the panel. Usually the difference between finding the cause in a minute and opening a terminal to hunt for it.
Security patches
Once a day the panel asks the system what patches are waiting and shows the count on the server card. Install them with a button, or let the server install the security ones by itself. It never restarts on its own.
Full root access
Your own OS, your own tools, your own cron. Everything the panel does — systemd, git, ssh — you can do by hand. The panel is a convenience, not a lock.
A console in the browser
The screen of the machine itself, not an SSH session — so a server you have locked yourself out of is still reachable. Useful exactly on the day the firewall rule turns out to be wrong.
A clean start whenever
Reinstall the system from the panel and pick a different image — eleven to choose from, from a bare Debian to Ubuntu with a hosting control panel. The address and the plan stay the same.
What is charged, and when
Once an hour, from the balance. Four states, four different amounts.
Running
The full hourly rate of the plan — the figure on the card above. An hour is the unit: a server taken for an evening costs an evening.
Stopped by you
A stopped server still holds its place on the drive and its IP address, so those two keep being charged — the amount is on the card and in the table. Processor and memory are free. This is what makes a test environment cheap to keep around.
Out of balance
When the balance will not cover the next hour, the server is stopped and the address is released. Nothing is deleted. Top up, start it again — the data is where you left it.
Archived
After 30 days without being started, a copy of the whole server is taken and the machine is removed from the node. Nothing is charged for the archive, and you restore it from the panel when you need it.
Paid on top
Two things cost extra, and both are optional.
Network copies of the whole server
A copy of the entire machine, taken on a schedule and kept outside the node — so a server that burns down is restored from a different drive. Snapshots are included in the plan, but a snapshot lives on the same drive as the server: it saves you from a bad release, not from a dead disk.
- Project S — €3 / mo
- Project M — €4 / mo
- Project L — €5 / mo
- Project XL — €6 / mo
Network storage
A separate disk of up to 20 TB, reachable over SFTP, FTP, Samba and WebDAV, which the panel can mount straight into a folder on your server. For media, archives and copies you do not want on an NVMe drive.
Full specification
Everything not in this table is the same on every plan.
| Spec | Project S | Project M | Project L | Project XL |
|---|---|---|---|---|
| vCPU | 2 | 4 | 6 | 8 |
| Memory | 4 GB | 8 GB | 12 GB | 16 GB |
| NVMe storage | 50 GB | 80 GB | 100 GB | 120 GB |
| Snapshots on the node | 1 | 1 | 2 | 2 |
| Network copies | €3 / mo | €4 / mo | €5 / mo | €6 / mo |
| Price per hour | €0.0205 | €0.0329 | €0.0438 | €0.0548 |
| Running around the clock | €15.00 / mo | €24.00 / mo | €32.00 / mo | €40.00 / mo |
| Stopped for a whole month | €9.00 / mo | €12.00 / mo | €14.00 / mo | €16.00 / mo |
| Team members | No limit | |||
| Traffic | Not metered | |||
| Region | Finland, Germany | |||
Questions
Can I change the plan later?
Yes, from the panel, in about a minute. Memory and processor change at once; the disk grows to the new size and the file system expands on the next boot. A disk cannot be made smaller — to move to a smaller plan, take a server on it and copy the data across.
What happens when the balance runs out?
The server is stopped and only the disk keeps being charged — nothing is deleted. Top up and start it again. After 30 days without being started, a copy is taken and the server is removed from the node; you can restore it from the panel.
Do I have to use the built-in git server?
No. The server is yours with full root access — deploy by hand, use GitHub, or install any tool you like. The built-in one simply saves you registering somewhere else and wiring up keys.
Where are the servers?
Our own hardware in two Hetzner data centres: Helsinki in Finland and Falkenstein in Germany. AMD Ryzen, NVMe drives, ECC memory on the larger node.
Who keeps the operating system up to date?
The panel asks the system once a day what patches are waiting, and shows the answer on the server card. Installing them is your call: press the button, or switch on self-installing security patches and the server handles them from then on. It never restarts itself — it only says when a restart is needed.
Can I just take a server, without any of the rest?
Yes. Pick a size and an OS, get root access in under a minute and do whatever you do — no repository, no deploys, no project. You can switch the rest on later, or never.
Start with one project
A server in under a minute, a repository in one click. Stop the server and billing stops with it.