GPU instances

DigitalOcean GPU Droplets vs RunPod

Choose DigitalOcean GPU Droplets when you want a posted hourly rate and per-second billing on a named card. Choose RunPod when a community-cloud pod or a serverless worker should match a short job, and you can accept that the rate depends on the GPU.

DigitalOcean GPU Droplets

On-demand GPUs in the same account as Droplets

From $0.76/hr

Strengths

  • The RTX 4000 Ada plan is a published $0.76 per GPU-hour
  • Billing is per second after a short minimum

Limits

  • An on-demand H100 is $4.41 per GPU-hour
  • Spot GPUs can be reclaimed

Vendor pricing Alternatives

RunPod

GPU pods and serverless inference, priced by the hour

Hourly, by GPU

Strengths

  • Community Cloud is usually cheaper than a single-tenant GPU
  • Serverless workers fit bursty inference

Limits

  • Community Cloud shares machines and the rate moves with the card
  • Secure Cloud costs more than Community Cloud for the same GPU

Vendor pricing Alternatives