Hi Arvon and all,
No offense taken: it's not my site!!  Although I don't agree with your assessment: lots of info presented with a nice web design.
No testing is possible, since the guy who created it doesn't have access to the unit yet: they just shipped last week.  That's why he states he's comparing specs, only.
As far as COSM: yeah, I agree it is gimmicky.  I have it on my V-Drums, and find it easy to use: that's the main advantage.  I haven't heard the amp modeling applied to guitar or bass samples, yet.
As for the 128 polyphony: the 5080 and XV-88 both have four-level velocity switching: most performances use 4 different patches, not layered, but velocity-switched.  Therefore, polyphony is not reduced, like you suggest, since you're just hearing one patch at any given time.  That's how many of the SRX pianos are set: four samples taken at four velocity levels, and not just a filter opening up to make it seem louder (like many synths do).
You can layer four sounds on the XV's as you mention, but there's only so far you can go with stacking sounds like that: velocity-switching sounds much more realistic, if set up properly.
Re: "timing problems": I've got a Roland A-90EX (with VE-RD1) that has problems with layering internal sounds: thick 4-zoned performances played with cluster chords sound like mush, due to sluggish timing.  (All fingers hit keys simultaneously, but an arpeggio comes out  

.  I didn't get this while demo'ing the XV-88 with thick performances.  Why?  Faster internal processing, I'd guess.
We'll all have to wait and see how good or bad the XV-5080 is: it's all just conjecture, at this point.  (Besides, none of this has anything to do with making music, anywway)   

Regards,
Chris