Jobs submit twice as fast and results land seconds sooner
A full audit of the job pipeline cut submission latency roughly in half and closed the gap between a runner finishing and the result being visible.
We audited every step a job takes from POST /jobs to a visible result and removed the waste. Submission latency dropped by about half (p50 from roughly 350ms to 150ms) by cutting database round trips out of the hot path. The gap between a runner finishing and the result showing up in the dashboard and API shrank by several seconds: the completion callback now publishes the result inline instead of waiting on a queue hop.
Same jobs, same outputs, less waiting. No API changes and nothing to update on your side.
Billing settlement was tightened along the way: every terminal job now settles from the runner’s own measured usage, with the price and its evidence recorded together.
The states a job moves through are documented in job lifecycle. An earlier change took settlement off the critical path, see faster job results.