On-Demand GPUs
Rent H100, H200, A100, and B200 GPUs the moment you need them, billed per second, with no contract and no minimum term.
What "on-demand" means on Aquanode
Every GPU listed in Aquanode's marketplace is available on-demand: no waitlist, no procurement call, no signed contract. You pick a GPU, deploy it, and it's billed by the second for as long as it's running. There's no minimum commitment period.
Instead of terminating a pod when you're not using it, you can stop it. A stopped pod stops billing and keeps your environment and data, so you can start it again later on the same or a different GPU without paying for the idle time in between.
Because Aquanode aggregates on-demand inventory from multiple independent GPU providers into one marketplace, you compare real, live prices for the same GPU model across providers before you deploy, instead of committing to whichever single cloud you already have an account with.
Cheapest live on-demand rates
| GPU | Lowest price/hr per GPU |
|---|---|
| H100 | $2.19 |
| H200 | $3.95 |
| A100 | $0.66 |
| B200 | $8.79 |
Live figures, refreshed hourly. See the full GPU Availability Index for methodology.
FAQ
Is there a minimum commitment for Aquanode's on-demand GPUs?
No. There's no minimum term or contract. Deploy, run, and stop whenever you want, and you're only charged for the time an instance is actually running.
How is on-demand GPU time billed?
Usage is metered per second while an instance is running. Stopping a pod instead of terminating it stops billing for the stopped time.
Can I stop an on-demand GPU instead of destroying it?
Yes. Stop instead of tearing a pod down between runs, then start it again later with your environment and data preserved.
How many providers does Aquanode's on-demand marketplace pull from?
Aquanode aggregates live on-demand GPU inventory from 11 providers across 62 regions into a single marketplace, so you can compare and deploy without opening separate accounts.
Related reading: H100 pricing and specs, H100 vs H200, A100 vs H100, H100 and H200: SXM vs NVL vs PCIe, and The best GPUs for AI, ranked.
