This is tricky to answer because I never compared the exactly same preamp with flying frontend and no caps to how it sounds with flying disabled and caps added. And because it used some sort of level shifting between the gain stage and the diff-amp (to avoid caps in the direct signal path altogether) several impedances had to be higher than I'd normally make them (to reduce power dissipation in the resistors with an extra up to 48V DC offset). So noise performance (at least at low gains) was worse than in a traditional preamp.
The advantage I saw was that you could switch phantom power on and off without any audible plops and clicks (unless the microphone made them). Also there's no rising noise at LF. But of course you still get LF rolloff, which is something most people overlook if they hear "no caps in the direct signal path". I'm not so sure whether I noticed any improved sound due to "no capacitors", since obviously there are still caps in the microphone itself. And those are usually the worst caps anyway due to the small space available inside the mic body. - Removing the cap in series with Rgain is a far greater improvement.
I built that thing almost 10 years ago, before I came up with the new servo circuit. Maybe I should try again... but wondering whether I should try again made me think whether the selected preamp circuit is the best one to use here.
Now with the new servo I simply use (non-electrolytic) caps in the input and place the Fc very low. Previously there was always the need to have it higher in order to have it not interact with the servo's moving Fc. With 22u caps I can get the 1/f noise corner to below 20Hz, so even the LF noise contribution of the caps can be ignored.
But still, being able to say "no caps in the direct signal path" is a nice thing to have (and the phantom blocking caps plus the caps in the microphones are currently the only caps left in my signal chain's direct signal path). Just that I'd only do it again after coming up with a preamp circuit hat doesn't move the DC conditions on the input around.