Should I look at GPU price per month or per hour?
Compare by the hour when your usage is bursty or uncertain, and by the month when you will keep a GPU running most of the time. A monthly or reserved rate is cheaper per hour but you pay for every hour whether you use it or not. The two only line up if you divide the monthly price by the hours you actually expect to run.
Hourly pricing bills only for the time an instance is up, so it fits experiments, spiky inference, and anything you start and stop. The per-hour rate is higher, but you waste nothing during idle time.
Monthly and reserved commitments trade flexibility for a lower effective rate. If you keep a card busy around the clock, dividing the monthly price across roughly 730 hours can beat the on-demand rate by a wide margin. If the card sits idle half the month, that discount evaporates and hourly would have been cheaper.
The honest way to compare is to convert everything to an effective cost per hour based on your real utilization. A monthly quote divided by the hours you will use is the number to put next to an on-demand rate, not the raw monthly figure.
Related questions
- Is renting a GPU cheaper than buying one?
- What is a GPU hour?
- What is the difference between serverless GPU and instance pricing?
Numbers on this page come from today's verified snapshot. Full table on the homepage; method in the methodology.