I do plan to run 1-2 VM at a time, so the 16gb is there to stay I'd say.
After doing research, the only difference I see between the two are:
0.2ghz speed difference
4mb l3 cache instead of 3mb l3 cache.
Could you tell me when one would benefit from these changes? Would it make compiling code any faster?
Absolutely not.
Oh, and if you assign RAM properly to your VMs, even 8GB should do you fine. But I guess you'll be more comfortable with 16GB, so this one's your call.
I normally assign only 2-4GB to each VM, and I only run 3 VMs at most on my 13" rMBP, because it only has 4 virtual cores.
On my 15" rMBP with 8 virtual cores, I normally run 6 VMs simultaneously, leaving 2 virtual cores for OS X itself.