There are a few things. Because I do so many other things with various text fields on my machine, and vim bindings everywhere is not really feasible (yet?!), being very adept with the Command/Option/Shift/Arrow Keys/Cut/Paste is something that has widespread utility.
The vim bindings for this stuff are much richer, and fewer key strokes of course, but the universality matters, and the buffer/clipboard dichotomy is pretty well handled in MacVim, but it's still the type of thing that is quite difficult to really dial-in without building for a specific OS from the ground up.
And of course there are a few nice things that developing for a specific GUI affords you where vim even with plugins isn't quite as elegant.
However that said, all editors involve tradeoffs, and Vim's ubiquity and raw editing power make it kind of a sure bet until the day we are dictating whole programs to orders-of-magnitude-smarter-than-human AIs. Emacs also gets the nod of course, since there is no chance of either of those going away, and no chance that anyone invents an editor that blows both of those away for all general purpose editing.
But I do have to give Allan Odgaard special thanks for forcibly showing me the light: that no matter how brilliant proprietary software may be, if it's my primary tool that I intend to use for the next 50+ years, closed source or a project with a risk of losing critical mass is simply a poor bet over time.
And of course I will still buy TM 2 when it comes out if he reverses his decision not to charge for it (which he should, and maybe give it away to individuals that complain, because who shouldn't be paid for years of secret toil?), and I'll likely still use it for its diff syntax highlighting, complex regex ack (because how long does it take to figure out the number of backslashes to escape something through ex, bash, and ack?). But it'll join BBEdit (which I use for encoding munging) in my quiver of specialized tools.