Cloud GPUs
GPU cloud pricing
Hourly GPU prices across providers, grouped by capability tier. Building an AI app? See the full stack →
| GPU | VRAM | Providers | Availability | Price range |
|---|---|---|---|---|
| NVIDIA H100 PCIe3 offers | 80 GB | Available | $0.99–$2.89/hr | |
| NVIDIA H100 SXM5 offers | 80 GB | Available | $1.49–$3.29/hr | |
| AMD Instinct MI300X1 offers | 192 GB | Available | $2.39/hr | |
| NVIDIA H2003 offers | 141 GB | Available | $3.59–$4.59/hr | |
| NVIDIA B2002 offers | 192 GB | Available | $5.98–$6.79/hr | |
| B3002 offers | 288 GB | Available | $6.94–$7.39/hr |
Frequently asked questions
How does DeployCost compare GPU cloud pricing?+
We index on-demand and spot GPU prices across providers, group them by model and capability tier, and show the full price range so you can spot the cheapest credible option — not just one headline rate.
What is the cheapest cloud GPU right now?+
Currently the cheapest indexed GPU is the RTX 5070, from $0.01/hr. Budget cards (T4, RTX 3060) are typically the lowest cost per hour.
What's the difference between on-demand and spot GPU pricing?+
On-demand runs uninterrupted at a fixed rate. Spot is much cheaper but can be reclaimed at any time — great for fault-tolerant training and batch inference, risky for live serving.