Careful reading of WD statements as well as testing suggests that two things are in play.
#1 Whatever the full operational rpm -- these drive goes to a reduced rotational speed (sleep mode) when not in use for significant periods (probably only 2-3 minutes) independent of OS power configuration. OS sleep mode still takes the drive to minimum (0) rpm and maximum energy conservation without respect for immediate performance once triggered. Any intellipower conservation speed above 0 rpm will of course still reduce the sustained disc spin current yet also reduce the time and current surge necessary to spin up to operational speeds. In fact the drive may reduce speeds in several steps over a relatively short time and the projected probability of operational use falls. There is probably a final optimal min/max for energy for maintaining long term sleep mode and energy needed to restore operational rotation speeds in a reasonable time depending on platter inertia, bearing friction, and aerodynamic resistances.
#2 Intellipower drives rely heavily on the large cache buffer to hide partial power downs and otherwise give the appearance of performance regardless of operational rotational speed. Specifically I suspect the WD references to sequential reads & writes being 7200 rpm drive equivalent means cache buffering speed dominants for most modest sized files when using simple capacitor backed write back and for reads data is predictively prefetched (assumed sequential) into otherwise unused cache space. Not sure but I thought lots of modern drives already did this if not turned off. Perhaps advanced capacitor backup of onboard cache and firmware make WD green cache so reliable such that it is never disabled even by OS request.
Thus WD statements about sequential file performance should be mostly true given that most computer-software does not frequently load large files over 30-64MB in an uninterrupted max speed stream even when drives can keep up. In fact the largest files are most often media files with rather low maximum flow rates (unless mastering movies from several sources of course) or databases where access is segmented for normal processing. Still the solution is limited by the 64MB cache window and the assumption of one primary task requiring sequential HD access. So all bets are off for more random file access patterns including a mixture of several different heavy file using processes. Most significantly in the face of a sustained system wide file demand the prefetched read data in the cache will be exhausted and write data will back up such that performance will drop to correspond to whatever the actual operational rotational speed is: 4500, 4900, 5400, 7200 etc rpm. I note that even 7200, 10K and 15K rpm drives will probably experience significant drops in perceived performance once the cache fills up with write data -- just not as much as slower green drives.