Billing is calculated by the millisecond. Provisioning and booting are not billed compute states. Compute billing begins when the instance enters running or rescue. A customer-requested shutdown can remain billable when the supplier keeps the GPU reserved; provider-side stops and certain suspensions are excluded. Ending the rental stops compute billing when the instance enters terminating, without waiting for destruction to finish. Storage is separate: any previously created billable volume continues until its own billing ends. New persistent volumes are not offered. Open Instances (/instances) to review the instance. See the existing “How are instances billed?” article (instance-billing) for the rest.
When does instance billing start and stop?
From the EverXCloud help center.
