I had a similar issue in NINA. In that case both NINA and APCC showed a time to meridian and they did not agree (by 2-3 minutes).
In my case the issue was a mount sync issue. I had NINA set to not sync APCC, and APCC had a polar align but I had done nothing in APPM to sync it (was not using a model either). So it was tracking nicely but was rotationally off a bit in where it thought it was (and a plate solving center on NINA had the object in view so I was clueless).
Allowing NINA to send a sync (which ASCOM or APCC reinterprets as a recal I think) fixed it and all match. I also am getting into the habit (as it might flush other issues) even if not using a model to go into APPM and do a plate solve and recal first thing after polar align in the evening.
Since that, the meridian timing is within a second or few, and flips work flawlessly.
Just a thought. The meridian screen in APPC shows time of meridian, it should match the transit time of the object (if SGP doesn't show it, you can look up in stellerium or others) . Unfortunately it stops showing it once it hits the meridian so it is tough to debug after it fails.