Vast.ai GPU pricing 2026
Marketplace that aggregates GPU capacity from independent operators worldwide. Real-time bidding model produces the lowest published H100 rates in the category at the cost of interruptibility and host variance.
Published rate card
Per-GPU per-hour rates pulled from the vendor pricing page linked above, Last verified June 2026. Rates exclude storage, egress, and any managed-service uplift. Reserved-capacity contracts typically improve on these rates.
| GPU | Configuration | Per GPU per hour |
|---|---|---|
| H100 SXM | Marketplace floor, interruptible | $1.490 |
| A100 80GB | Marketplace floor | $0.790 |
| L40S | Marketplace median | $0.890 |
| RTX 4090 | Marketplace median | $0.450 |
| RTX 3090 | Marketplace floor | $0.210 |
Hidden costs to watch
- Listings are individual hosts; reliability, network speed, and host trust score vary widely.
- Interruptible instances offer the lowest rates but can be reclaimed at any time.
- Data transfer in and out is metered by the host's network arrangement, not Vast.
What Vast.ai is best for
Cost-sensitive ML hobbyists, research, and stateless inference where interruptibility is acceptable.
See the GPU cloud buying guideWorked example
Acme Vision Co. (illustrative example, not a real company) needs to train a 10 billion-parameter vision-language model on a fixed 8-GPU cluster for 30 days at 18 hours per day duty cycle.
At Vast.ai's cheapest published rate of $1.49 per GPU-hour, the run costs $6,437 for raw GPU compute, before storage, egress, and MLOps overhead. Add the typical 25 percent year-one uplift and the modelled spend is $8,046. Use the calculator on the homepage to model your own GPU class, cluster size, and duty cycle.
Last verified June 2026. Vast.ai rates change frequently. Always obtain a vendor quote before purchase.
Visit Vast.ai pricing page