At least one of your assumptions must be wrong. I'd guess
that
Pads consists only of square made of metal layers.
is one of them. Maybe you want to eyeball the netlists and
see just what kind of parasitics are thrown down.
But why pad parasitics would make that much of a problem,
is a question. Is amplifier drive extremely weak? Test
bench (input / feedback elements) so high-impedance that
pad parasitics somehow whack the operating point, or
the way you are picking up gain/phase numbers maybe
"works for ideal, not so good for nonideal" (like, assuming
AIN is the source amplitude and not the pin difference
voltage)?