OPamps still inferior to discrete circuits?

There are advantages of both. Discrete circuits can be made for some special characteristics (such as very high voltage or lowest possible noise for some given source impedance), or customized with features that just don't have enough of a market to interest IC designers.

But there are some things that can be done in integrated circuits that just can't be done in fully discrete circuits, such as extremely high inherent transistor matching and highest possible thermal coupling between devices that need to track to work right.

So far in opamps the best I have seen for audio is made by Dan Kennedy:

www.diyfactory.com/data/Great_River_Opamp2.pdf

It is built up from discreet components.

Just to add to this old thread, I just replaced the OPA827 opamps in the output stage of a low frequency EQ unit that fits between my preamp and my amps, with the new Burson 4th generation discrete opamps (on sale for $68 per pair at PCX). The circuit is a unity gain output stage, and the entire music signal passes through it on its way to the amps.

Wow, the difference is very obvious. Deeper and cleaner bass, and complete removal of the last traces of hardness in the treble. Very smooth sound, but retaining all the detail. The OPA827 was the best IC opamp I'd heard in that position, but the Burson is definitely better.

The Burson operates with 20mA idle current vs the OPA827's 5mA, so that may account for the improved bass drive. I'm impressed.