From what I see, to do a decent job you'll be learning EM
solver tools and often you end up with a solution you did
not want - like for a digital I/O, do you really want a s2p
model like RF dudes (who do most inductor work) prefer?
And on the flip side, is a lumped element model (of arbitrary
lumpiness) going to do you any good?
And when you start caring about Q, you start caring about
the substrate and all that. People seem to make careers of
integrated inductor modeling.
I imagine that you might be able to find papers on 0.13um
inductors (from some while back when people cared about
0.13um technology and you could get published), and get
dimensions / SRF / Q info and so on. You'd likely have some
work left to do, to get anything that worked for time-domain
simulations, but maybe you will get lucky and find smaller
geometry examples.
On the other hand, small inductors are, well, small and may
simply not give you the value you want. If the value you
need is approaching the value of the "too big" inductor
then "too big" may be the end of the story.
Also, a bond wire can make a good enough inductor if
you happen to want single-nH-range, or can design
such that this works out OK. People do make use of
this "feature". You may not be doing a wirebond part,
of course.