Good point. Differences between direct notebook execution and API-triggered runs can often come from session initialization, environment variables, cluster warm-up time, or request timeouts.I’d compare the runtime environment and timestamps for both ...
