Pay for What You Use
No rate cards to go stale. Your exact price is calculated live in the console when you deploy — you always see it before you commit.
Compute
Virtual servers and bare metal with dedicated CPU and RAM.
Per-second billing, 1-hour minimum
GPU Instances
NVIDIA A100 and H100 instances for AI/ML training, inference, and rendering.
Priced per GPU-hour, quoted live
Storage
NVMe block storage, snapshots, and backups.
Metered per GB-month
Network Transfer
Generous outbound transfer included with every server.
20 TB included per server
Managed Kubernetes
Production-ready K8s clusters. Control plane managed, you own the workloads.
You pay for the nodes
Support Plans
From community support to dedicated engineers with custom SLAs.
Community tier included
How pricing works
Per-second billing
Compute is billed per second with a 1-hour minimum. Stop an instance, stop the meter.
Live market pricing
Rates are calculated at deployment time from public market data. The console shows the exact price for your config and region before you commit — no stale rate cards.
Commit to save
Monthly commitments save ~30% over hourly; yearly ~55%. GPU and bare metal plans can be reserved for further discounts.
What you can deploy
The console shows you the exact live rate for any of these before you deploy.
| Family | Configurations | Notes |
|---|---|---|
| Virtual Machines | vm-standard (4 vCPU / 16 GB / 80 GB NVMe) | Per-second billing. Linux, Windows, and custom images. |
| Bare Metal | m4.metal.small (6 cores / 64 GB) → rs4.metal.xlarge (64 cores / 1.5 TB) | Single-tenant dedicated hardware. No noisy neighbors. |
| GPU Instances | 1×–8× A100 40GB, 1×–8× H100 80GB | Pre-configured ML images: PyTorch, TensorFlow, CUDA 12. |
| Storage & Networking | Block storage, snapshots, Elastic IPs, load balancers | Block storage and snapshots metered per GB-month. 20 TB egress included per server; private networking is free. |
Full technical catalog in the billing docs.
High-volume or reserved-capacity commitments? Volume discounts are available — talk to us.