I've run a 396 and XT side by side for about two years now. I've used this arrangement in two rallies and numerous solo rides with multiple waypoints.
The XT has a far better screen, but in most other regards I find the 396 actually performs better in the real world. Primarily the 396 is much, much better at dynamically re-routing when I go off route for any reason. The 396 always recalculates faster and more reliably. The XT sometimes gets stuck in a "do a U-turn" loop even when I'm back on the main route, forcing me to stop and restart the route. The 396 never has this problem.
I have both loaded with the same maps, configured with the same settings. It's infuriating and puzzling. Maybe all the extra features (whatever they are) bog down the XT's CPU? I don't know.
If I had to replace my XT tomorrow I would get another 396, even if they were the same price.