Doubt about gps accuracy

With rtk systems, I know you either need a fixed base that spent time surveying, a fixed base with accurate coordinates for it’s location or a ntrip system, to feed corrections. In the moving base setup or gps for yaw that you use two gps modules in the same vehicle, with the moving base giving rtcm corrections with the addition of reference station information so the rover can estimate it’s heading with precision.
Disregarding the gps for yaw function, since there are two gps modules within the vehicle with a fixed distance between their antennas, does this setup have increased absolute accuracy in comparison if I just used just one gps module?If so, how much?

No. No accuracy increase compared to single GPS.

Agree with @amilcarlucas . Even when using Blend I don’t think there is a real improvement. You can have one with very good positional info (call this clear water) and the other with not as good (call this muddy water). The blended result is muddy water…
For data redundancy in case of failure it has value.

Assuming that you have two perfectly working rover GPS, both receiving the same phase correction data from the same base. It means that they will be identically accurate. (Theoretically)

RTK rover precision is ALWAYS relative to the BASE. So if the base is off, they will be off as well, regardless of how accurate they are relatively, blending them will not increase the precision.

But in reality, gps receivers are never the same, they can differently react to temperature, noise, power etc… So at the end, you will have only muddy water.