Loading…
Loading…
Pay only for what you use. Automatic failover and load balancing included at no extra cost.
Perfect for development and variable workloads
Most popular
For production workloads with predictable usage
For large-scale deployments with custom requirements
Per GPU, per hour
All prices in USD. Reserved pricing requires a 1-year commitment.
| GPU Model | Pay as You Go | Reserved (1yr) | Savings |
|---|---|---|---|
| H200 | $2.49/hr | $1.74/hr | 30% |
| H100 | $1.80/hr | $1.26/hr | 30% |
| A100 | $0.80/hr | $0.56/hr | 30% |
| RTX 4090 | $0.29/hr | $0.20/hr | 30% |
| RTX 4080 | $0.39/hr | $0.27/hr | 30% |
| RTX 4070 Ti | $0.29/hr | $0.20/hr | 30% |
| RTX 3090 | $0.29/hr | $0.20/hr | 30% |
| RTX 3080 | $0.19/hr | $0.13/hr | 30% |
Need higher volume? Contact us for custom pricing.
We bill per second of GPU usage. You're only charged when your workloads are running. Failover to backup nodes is free—you pay the same rate regardless of which nodes serve your request.
Yes. Automatic failover and load balancing are included at no extra cost. When nodes go down, we route to healthy nodes instantly—same price.
Yes, you can upgrade or downgrade at any time. Pro plan discounts are applied immediately to all new usage.
Pay as You Go has no commitment and charges standard rates. Reserved capacity requires a 1-year commitment but offers 30% savings and guaranteed availability.