Pay only for the seconds you run. No setup costs, no egress surprises, no minimum commitments. Choose on-demand for guaranteed capacity or interruptible for maximum savings.
Dedicated, non-interruptible pods. Run until you stop them.
Spare capacity at a discount. May be reclaimed during peak demand.
| GPU model | VRAM | TFLOPS (FP16) | On-demand | Interruptible |
|---|---|---|---|---|
| NVIDIA H100 SXM | 80 GB | 1,979 | $2.49/hr | $1.39/hr |
| NVIDIA A100 | 80 GB | 624 | $1.29/hr | $0.69/hr |
| NVIDIA RTX 4090 | 24 GB | 330 | $0.58/hr | $0.23/hr |
| NVIDIA RTX 3090 | 24 GB | 142 | $0.31/hr | $0.14/hr |
Billing
The moment your pod reaches the RUNNING state, the meter starts. The moment you stop or terminate it, the meter stops — down to the second. No rounding up to the nearest hour, no minimum runtime.
Billed for exactly 2,242 seconds — not a full hour.
Your funds
Add credit to your prepaid balance via card or wire through our PCI-DSS compliant processor. Poidex never stores raw card numbers.
Your balance is held as account credit and is never invested or commingled with operating funds. Unused balance remains yours and is refundable.
Charges are metered per second only while a pod runs or a volume is provisioned. Every deduction is itemized in your billing dashboard.
Unused prepaid balance is fully refundable within 14 days of deposit. Request a refund of any remaining balance at any time — see our Refund Policy.