The WASAPI exclusive mode directly renders the audiostream of an application (osu! in this case) to the audio adapter of your PC/Laptop and therefore has lower lantency. Another neat effect of using the exclusive mode is, that only the audio from osu! will be heard and no background sounds from other applications (this is good if you want to focus on the game and don't want to be disturbed).
If implemented, please make it an option to use the exclusive mode (the normal mode aka shared mode should be the same as now). Because some people maybe want to hear sounds from other applications and the delay will be not the same as it is at the moment (or you create a third option which uses about the same buffersize as it is now, but you won't have background noise).

Three output options:
- WASAPI exclusive mode
- WASAPI shared mode
- WASAPI exclusive mode with about the same delay as the current build (if you are already used to the current delay)

EDIT:
How audio output currently works in osu! (my assumptions):
Certain small amounts of audio are renderd to the windows audio mixer consecutively, while the rest is buffered and kept ready to be mixed with a hitsound if you press a note/slider/whatever. Therefore you press notes earlier than they are actually heard, to match the delay and to hit accurate. With the WASAPI exclusive mode the delay between the keypress and when you actually hear your hitsound is lowered a bit and tapping should feel a bit more natural as it is more accurate. Mind you that it is only one part of the processing chain (mixing hitsounds with the source audio and delay of your input/output device are other parts of it).

EDIT 2:
A downside of this would be that not all soundcards support the WASAPI exclusive mode.

WASAPI is a Windows specific API. Peppy is trying to go for full cross-platform compatibility (hence the recent OpenGL stuff), so consider this idea rejected.