Sizes

A size is chosen when a resource is created and describes what one instance of it is provisioned with. Sizes and plans are independent: a plan decides what an organization may do, a size decides what one resource gets.

Web services

Every size scales to zero: a service with no traffic runs no instance and costs nothing for compute until a request arrives. The upper bound is how many instances Kuppit will run at once under load. A new service is Starter.

Workers

A worker's size is what each of its instances gets; how many instances it keeps is its own setting, from zero to ten, and is billed for as long as they are allocated. A scheduled worker's size is what each run gets, billed only while the run executes. A new one is Starter.

PostgreSQL

A size is a database's compute ceiling. It scales down while idle and suspends after a few idle minutes, waking on the next connection. Storage is not part of the size: it grows with the data stored and is metered as used. A new database is Starter, with PostgreSQL 18.

Costs

Infrastructure is billed on usage, at published rates: compute and memory while a service is serving, while a worker's instances are allocated and while a scheduled worker's run executes, and a database's compute while active and storage while it exists. A service that scales to zero costs nothing while idle.

When you choose a size, the dashboard shows about $X/month for it. That figure is the rates applied to a stated month — one instance running the whole month for a web service or a worker, thirty runtime hours for a scheduled worker, storage alone for a database — and the sentence beside it says which. It is for comparing sizes, not a price: the bill is what actually ran, and the dashboard's Usage page shows it as the month goes.