http://www.tonegym.co/space/talk?id=SCH01ZAPDBM&t=VU4FHM1LOSQ

@Yossi Patt has asked about this a couple days ago. I copied the link to that post above, for you to find my extensive explanation for Rhythmaniac.
You also should always measure and adjust your input delay deficit with the latency compensation settings right here :
https://www.tonegym.co/account/latency
Depending on processing, the delay can change noticeably between games, stages or sometimes even attempts if one happens to be that unlucky. The auto latency tool sometimes work, but not always for me, so when in doubt, manually test and set it accordingly.
Remember that the auto tool as well as default options are only consider input latency and !not the output latency nor both as that is your true delay,
Other than putting the right settings, I would assume that your problem doesn't seem to the latency delay, but rather the amount of hits you do opposed to the exact same number as requested by the a stage.
If you hit exactly as many times as the question asks for, even with all hits within the first beat on the first bar, so regardless of timing, both games will show both the levels and your live input recorded rhythms in comparison and always will state : Inaccurate!
I can't tell you the exact reason since I don't see you playing nor inside your brain while you do it, but try to count the hits of both the example and your input, or the amount of notes and your input, whichever of those 2 games you're playing.
If you still feel you get Insufficient tabs, even with a big buffer pause at the last Beat and 4 hits in 4 bars are too blatantly countable, maybe you start in too early... just an option.
The easiest way to visualize your current latency is matching the amount of hits while emphasizing the metronome hits and see how accurate they are.

On my system it's extremely accurate with no delay, so that's what you can aim for.

Great, @Jason Blank 🙌 But unfortunately not everyone has the luxury of a 0 latency system 😌