DigitalOcean GPU Droplets GPU pricing 2026
DigitalOcean integrated GPU compute into the Droplet control plane in 2024. H100 / H200 / MI300X / MI325X options with a strong long-term-reserved pricing tier and the Paperspace/Gradient inheritance for ML tooling. On-demand rates were restructured upward on 1 August 2026 while reserved rates held, widening the reservation discount.
Published rate card
Per-GPU per-hour rates pulled from the vendor pricing page linked above, Last verified July 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 80GB (HGX) | On-demand, per-GPU (1 or 8x) | $4.410 |
| H100 80GB (HGX) | 12-month reserved, per-GPU | $3.260 |
| H200 80GB (HGX) | On-demand, per-GPU (1 or 8x) | $4.470 |
| H200 80GB (HGX) | 12-month reserved, per-GPU | $3.400 |
| AMD MI300X | On-demand, per-GPU (1 or 8x) | $2.590 |
| AMD MI300X | 12-month reserved, per-GPU | $1.910 |
| AMD MI325X | On-demand, per-GPU (1 or 8x) | $3.800 |
| L40S | On-demand, single GPU | $1.570 |
| RTX 6000 Ada | On-demand, single GPU | $1.570 |
| RTX 4000 Ada | On-demand, single GPU | $0.760 |
Hidden costs to watch
- The 1 August 2026 restructure widened the reserved discount: H100 on-demand rose to $4.41/GPU-hr while 12-month reserved held at $3.26 (about 26% off), so a commitment now saves meaningfully rather than modestly.
- Block storage, snapshots, and bandwidth metered separately.
- Capacity windows for H100 / H200 are constrained; provisioning lead time is a real planning factor.
What DigitalOcean GPU Droplets is best for
Existing DigitalOcean customers who want H100 / H200 access on the same control plane as their Droplets and managed databases.
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 DigitalOcean GPU Droplets's cheapest published rate of $0.76 per GPU-hour, the run costs $3,283 for raw GPU compute, before storage, egress, and MLOps overhead. Add the typical 25 percent year-one uplift and the modelled spend is $4,104. Use the calculator on the homepage to model your own GPU class, cluster size, and duty cycle.
Last verified July 2026. DigitalOcean GPU Droplets rates change frequently. Always obtain a vendor quote before purchase.
Visit DigitalOcean GPU Droplets pricing page