Per-second billing, explained

Every Nexpod product bills by the second: Cloud Servers, Serverless workers and Instant Clusters. Rates are quoted per hour because that is easier to compare, and the invoice divides them by 3,600. This post shows exactly what gets counted.

What the meter counts

• Cloud Servers from the moment the server reports ready until you stop or delete it.
• Serverless only while a worker handles a request or warms up. A pool at zero costs nothing.
• Instant Clusters per node, with the same start and stop rules as a single server.
• Volumes separately, for the space you keep, whether or not a server is attached.

A worked example

Say you start a B300 server for a fine-tuning job. The server rate is €48.00 per hour. The job ends after 47 minutes and 12 seconds and you stop the server straight away.

Rounded to the hour, the same job would cost €48.00. On one job the gap is small. For a team that starts dozens of short experiments a day, it decides whether you pay for compute or for idle time.

Stopping or deleting

Stopping a server ends compute billing and keeps the attached volume. Deleting it also hands the hardware back. If you plan to return within the hour, stopping is enough: your data stays put.

Reading the invoice

Invoices arrive monthly, in euros, with VAT on a line of its own. Each line item shows the resource, its region, the seconds billed and the hourly rate, so any row can be checked by hand. The same data is available from the API.

In short

You pay from ready to stop, by the second, in euros. Provisioning is free, idle Serverless pools are free, and stopped servers only cost what their volumes hold. Every rate is on the pricing page.