Two meters: server time per hour and agent work per token. Schedule your servers to save. All prices in USD. No subscriptions.
We pick the lowest-cost server that fits your specs.
* Monthly and yearly figures assume the server runs continuously. You are billed only while it runs.
Start with what your app does. Select a type to see every size and price.
Choose a region and server when you build.
Billed only while your app runs.
| Server | vCPU | Memory | GPU | Price / hr |
|---|
Agent work is charged per tool call via token counting.
A stop-work safeguard is available.
Plus a flat $0.05 service fee per prompt, charged once regardless of token usage. Foundation model used.
Hourly, only while your app is running. Schedule to save.
Per token, plus a flat $0.05 per prompt.
No seats, no tiers, no minimums. Pay for what you use.