I think my apexi 2.5 bar gauge is doing this too, as I can see what you mean. If it calibrates to zero when switched on, and the engine is obviously on vacuum at the time (cranking) it's going to over read by the vacuum amount??

I have a boost controller with boost display (profec B spec II) and it disagrees with the apexi by half a bar under at 2 bar!!

I have a mechanical gauge coming so will be able to check for sure shortly!!