The issue here is how you define the top x% of hands - and differences in how this decision is arrived at are likely causing you to see what appear to be different results when using different tools.
For example, the tool you have linked to in your question gives options to specify the number of opponents, so this implies it is determining the rankings of individual hands based upon their equity against that number of opponents. There are plenty of instances where a hand has a greater or lesser amount of relative equity depending on the number of opponents you are facing. This means that when using this approach, a different number of opponents in the hand will affect which specific hands might fall inside or outside of the top 5% of hands for example.
PokerStove is the tool I've used in the past for range and equity calculations, although I'm sure there might be better and more readily available tools out there now. If I ask PokerStove for the top 10% of hands (which it actually rounds down to 9.8% for me), I get:
88+, A9s+, KTs+, QTs+, AJo+, KQo
whereas your linked tool gives me:
66+, A9s+, KTs+, AJo+
for 1 opponent
77+, A9s+, KTs+, AJo+, KQo
for 2 opponents
77+, A9s+, KTs+, QJs, AJo+
for 3 opponents
which are all slightly different.
This illustrates that there are a variety of ways to arrive at an answer to the question "What is the top 5% of hands in holdem?" and different tools will take different approaches to this. It's therefore very difficult to say which one is "best" or which ones are "correct", as the differences are likely not due to any of them being "wrong", but rather just calculating something which is slightly different.