We’re seeing increased latency for jobs initiated via the agent API, and are currently investigating the underlying issue.