The first thing I’d check is whether both executions are actually running under the same conditions. API-triggered jobs can have different initialization timing, environment variables, dependencies, permissions, or cluster state compared with a notebook session. Capturing those details alongside timestamps and job parameters should make the source of the difference much easier to identify. The same principle applies when comparing mobile-based execution environments, where device and runtime differences can affect results mobile Roblox execution tools  are one example of that kind of environment-specific testing.