Compute pricing

Work runs on serverless functions. You pay for the CPU time your work actively uses, the memory provisioned while it runs, and each invocation, the same way the underlying cloud bills it.

  • API and MCP requests
  • Agent turns
  • Pipeline runs
  • Predictions
  • Scheduled and queued jobs

Prices

Live prices are briefly unavailable. They will be back shortly; in the meantime, contact us for a quote.

How it's measured

  • CPU time actively used while your work runs. Time spent waiting on the network or a database is not counted. When several pieces of work share a function, the time is split fairly between them.
  • Memory provisioned while your work runs, multiplied by how long it runs.
  • Each request or job that starts work.

Example workloads

These workloads illustrate what is measured. Cost calculations are available when live rates return.

A 30-second pipeline step

20 seconds of active CPU with 2 GB of memory for 30 seconds, in London.