Cost and billing
GPU-hour
What is gpu-hour?
A GPU-hour is one accelerator held for one hour, and it is the standard unit for compute pricing. It makes costs directly comparable across providers in a way per-token pricing does not, because token rates hide both the hardware and the utilisation.
Total cost is the hourly rate multiplied by wall time, which means a faster and more expensive class can be cheaper overall. Comparing hourly rates alone is comparing half the equation.
Billing granularity matters at small scale. Per-second billing above a short minimum is materially different from hourly rounding when a job takes eleven minutes.
Related terms
Start with the free tier
A magic link creates your account, your tenant and your first API key. No card until you ask for compute.