Bid on compute resources and pay market prices. Prices update in real-time based on supply and demand.
Choose a plan that fits your needs. All plans include auction-based compute pricing.
Perfect for learning and small projects. Pay only for compute usage.
For growing teams that need more power and better support.
For large organizations with advanced security and compliance needs.
Current market prices for GPU compute. Prices update every few seconds.
| Instance | GPUs | GPU Memory | vCPU | RAM | Current Bid | Change | Availability | Action |
|---|---|---|---|---|---|---|---|---|
| H200 8x SXM | 8x H200 | 1.1TB HBM3e | 224 | 2TB | $15.99 | +1.8% | MEDIUM | |
| H100 8x SXM | 8x H100 | 640GB HBM3 | 192 | 2TB | $7.99 | +2.3% | HIGH | |
| H100 8x PCIe | 8x H100 PCIe | 640GB HBM3 | 128 | 1TB | $6.49 | -1.1% | HIGH | |
| A100 8x 80GB | 8x A100 | 640GB HBM2e | 128 | 1TB | $5.99 | +0.8% | HIGH | |
| L40S 8x | 8x L40S | 384GB GDDR6 | 128 | 512GB | $3.99 | -2.1% | HIGH | |
| RTX 4090 8x | 8x RTX 4090 | 192GB GDDR6X | 128 | 512GB | $2.49 | -3.2% | HIGH |
Standard and memory-optimized CPU instances for general workloads.
| Instance | Type | vCPU | RAM | Storage | Current Bid | vs On-Demand |
|---|---|---|---|---|---|---|
| compute.16xlarge | Compute Optimized | 64 | 128GB | 2TB NVMe | $0.89 | -67% |
| compute.8xlarge | Compute Optimized | 32 | 64GB | 1TB NVMe | $0.45 | -64% |
| memory.16xlarge | Memory Optimized | 64 | 512GB | 2TB NVMe | $1.29 | -71% |
| general.8xlarge | General Purpose | 32 | 128GB | 1TB NVMe | $0.52 | -59% |
Transparent pricing for storage, networking, and support.
High-performance persistent storage for your clusters.
Generous free tiers with predictable overage costs.
Get help when you need it from Kubernetes experts.
Common questions about our auction-based pricing model.
You set a maximum bid price per hour for your instances. Our system matches your bid with available capacity. If the market price is below your bid, you pay the market price. You'll never pay more than your bid.
You can configure how your cluster handles this: gracefully scale down, maintain current capacity at your bid price (if already running), or fall back to on-demand pricing. Your workloads are never interrupted without warning.
No. You pay for compute (at auction prices), storage, and egress over 100GB. The platform fee covers the control plane, monitoring, and standard support. Enterprise plans have custom pricing.
Most customers save 50-80% compared to AWS, GCP, or Azure. The exact savings depend on your workload flexibility. Batch jobs that can wait for low prices save the most. Real-time workloads that bid higher still save 30-50%.
Create a free account and get $50 in compute credits. No credit card required.