I believe that:
The standard KGSS has a PS of +/-350 volts.
The BHSE has a PS of +/-400 volts.
The KGSSHV has a PS of up to +/-500 volts.
The KGSS Carbon has a PS of around +/-450 volts
The DIY T2 has a PS of +/- 500 volts
If we assume that the amps can swing up to their PS rails (not really, but relatively close) then:
The KGSS can swing 1400 V peak to peak
The BHSE can swing 1600 V peak to peak
The KGSS Carbon can swing up to 1800 V peak to peak
The KGSSHV and DIY T2 can swing up to 2000 V peak to peak
Now, that seems like a fairly large spread in voltage, however in decibel terms, the difference between the KGSS and KGSSHV/DIY T2 is about 3 dB in loudness. Remember that with most Stax headphones, 100 volts RMS (283 V peak to peak) is about 100 dB loudness, which is loud enough to give you permanent hearing damage with an 8 hour exposure, so a KGSS gives you about 112 dB dB, and the others go up from there. Also, when the peak to peak voltage exceeds twice the bias voltage, which is 580 volts for Stax, there is a chance of damaging the headphones as well.
The reason for using higher voltages (and currents) is not so much for increased loudness but for more headroom at normal listening levels so that the amp is just loafing. To some degree this notion also depends on how much the music has been compressed (see Wikipedia "loudness wars"). There has usually been some degree of compression use so that the softest music is above the noise level inherent in the recording process (e.g. tape noise), but acoustic music generally has had less of it. In the '80s to 200's, a fair amount of pop/rock music was compressed all to hell so that it was basically all one level (LOUD!) from beginning to end, but it appears that trend may be decreasing. If you are playing compressed recordings at loud levels, you may be pushing the amp toward its limits. Kevin Gilmore noted that at one meet, someone was playing his DIY T2 so loud that you could here it clipping across the room, since Stax headphones are open back.