Afraid you are quite wrong, both the diagrams actually show exactly how many opamps the x5 and dx90 use. The x5 uses a single dual channel opamp for LPF and another for AMP, while dx90 uses two single channel opamps for each, the dx90 is technically the better design, using two extra opamps over the x5 to completely separate the channels.
The reason why the x5 and dx90 have two dual channel opamps after the dac is because both dacs output a 2 channel current for each left and right channel (so 4 channel total), so either 2 dual channel opamps are needed, or 4 single channel opamps, after this the signal is converted to 2 channel and you need either 1 dual channel opamp or 2 single channel op amps the rest of the way down the chain.
No way to know that for sure so I didn't get too involved beyong a cursory check. The last 1612 of the X5 can't feed 2 channels from one output so it's impossible to know the exact chip count because the diagram doesn't actually make sense. The biggest difference is the balanced use of the DAC ind DAC itself of the DX50 along with the summing or differential input to single out in the LPF stage. If you think a balanced Sabre is better than a single 1792 it's better if not, not. Last time you said the 1612 is better than the 1602. Is one 1612 better than two 1602? The answer will always be, it depends.
There are the same number of stages doing the same functions. How it relates to performance is speculation. There is no better or worse in those 2 paths. It will come down to execution.