Skip to content
Coming soon

On-demand servers: a full Linux server in seconds

Launch a Linux VPS server whenever you need it and pay only for the time you use it. Billing is calculated by the hour, and you decide how long the server runs – set an automatic shutdown or keep it running until you no longer need it.

The service is still being prepared for launch. Orders are not yet being accepted.

Secondsfrom start to a booted server
< 30 stypical time to root SSH
1 hminimum billed session
IPv4public address with every session
Why on-demand

Instant servers, session pricing

Everything a real server gives you — for exactly as long as you need it.

A real server in seconds

On-demand servers boot from prepared Linux images on dedicated Firecracker nodes. The control plane starts the microVM in seconds, and root SSH is typically ready well under half a minute; exact readiness varies by image and network.

Pay per session, not per month

Modern session metering: billed by started hours, and you control the duration — set an auto-stop time yourself or keep the server until you delete it, with no imposed maximum. Billing ends with the session — no monthly plan, no forgotten invoices.

Your server, your rules

Full root access over SSH on your chosen Linux image. Every session includes one required public IPv4 address; install and test software within the service and network policies.

Zero cleanup afterwards

When the clock runs out the VM destroys itself completely — nothing keeps running, nothing keeps billing. Destroy earlier from the panel whenever you are done.

Use cases

What on-demand servers are for

Start a clean server when a build, test, demo, or experiment needs its own machine.

  • Try risky software safely

    Run that curl-pipe-bash installer or unknown binary on a disposable box instead of your production server.

  • AI agent sandboxes

    Give an AI agent a real machine with a built-in expiry — it can install tools and run code without touching anything permanent.

  • Build & test runners

    A clean, fast environment for one-off builds, integration tests, or reproducing a bug exactly from scratch.

  • Demos & workshops

    Spin up a machine per attendee or per demo, hand out SSH access, and let everything expire when the event ends.

  • Benchmarks & PoCs

    Measure a database, compile flags, or a proof-of-concept on consistent NVMe hardware — collect the numbers, then let it go.

  • Learning & interviews

    Practice Linux, teach a class, or run a technical interview on a real server where mistakes cost cents, not servers.

Pricing

Simple session pricing

Billed by started hours, minimum one hour. You see the price before you start; set an optional auto-stop or run for as long as you need.

  1. Boots in seconds
  2. Prepaid whole hours
  3. Auto-stop or delete anytime

Standard

For tryouts, sandboxes, and demos. The quote uses the current Standard plan rate and session multiplier, billed in whole hours (minimum one hour).

Performance

More vCPU and RAM for builds and benchmarks. The quote shows the current Performance resources and exact prepaid session total.

How the clock works

You choose the session length up front and prepay it. Destroy early whenever you finish — the session simply ends; nothing keeps running in the background.

Choose the right execution model

Functions and on-demand servers are one family of on-demand compute, and both sit under a single entry in your panel. Pick a function for short, bounded code and webhooks; pick an on-demand server when the job needs a whole Linux machine with SSH.

FAQ

Frequently asked questions

01

What is an on-demand server?

A full Firecracker Linux microVM you can SSH into — a real root environment, not a serverless function. Use it for minutes or days; billing is by whole hour. Optional automatic stop, or run until you delete it.

02

How fast does an on-demand server start?

In seconds. Prepared Linux images start quickly, and root SSH is typically ready well under half a minute. Exact readiness varies by image and network.

03

How is this different from Functions?

Functions execute bounded code and bill per invocation. An on-demand server provides a complete Linux environment with SSH and disk for work that needs an entire machine. Both use Firecracker isolation, but they are separate workflows.

04

What happens to my data?

The server's local disk is deleted with the server. Automatic backups and snapshots are not included. If you set an automatic end time, store important data elsewhere before that time; otherwise, do so before deleting the server.

05

Can I send email from an on-demand server?

To reduce abuse, on-demand servers restrict outbound connections to mail-delivery TCP ports 25, 2525, 465 and 587. This restriction cannot be removed for an on-demand server. This policy does not block incoming traffic, DNS, the service hostname or other outbound ports. Use an HTTPS mail API or a monthly VPS if the workload must send email.

06

Can I convert an on-demand server to a monthly Cloud VPS?

Sessions stay sessions — there is no in-place conversion. If the project outgrows the session, order a monthly Cloud VPS or Firecracker VPS from the same panel; it only takes a few minutes to move.

07

Is ordering available now?

Not yet. Customer ordering is disabled while VPSnet completes product and reliability validation. This page will be updated when ordering opens.