Its a bit weird that the sizes are different, but if the content is the same (and appears the same when you view the feature classes in the GDB) whats the concern?

How much is the difference anyway?

The main point is that our customer challenge it

difference is around several bytes to 1K in size.

This could be as simple as different timestamps on the files inside the zip being compressed differently.

I understand the concern, and if you want to be absolutely sure you can always compare the (unzipped!) files.

That was one of my thoughts. There may also be some level of randomness in the compression. One instance may use a slightly better seed/random number to perform the compression.

You also say you're using the same version, but are the machines the same? different versions of windows, with different specs may have different settings when compressing.

Here are default the settings from 7z on my laptop. Looking at another machine i have access to, the defaults are different because the specs are different

Basically, there are so many variables that are likely to play into this. If it were mbs different, then there could be an issue, but if the contents of the file are the same, then i don't see the problem with a couple of bytes difference.

It could even come down to the "pysical" position of the data on the drive. on one system the file might span two blocks and on the other system it might span three