The R5/R6 have DPAF II while I am pretty sure it is DPAF Mk I on the 1DX III. The difference is not something I have done a deep, or even shallow, dive into, but yes, there may be (likely is) a technical difference.

Nuances in the actual performance between PDAF and DPAF II is something I would be interested in but have not seen a good evaluation. What I would suspect is that you would see different strengths and weaknesses. Personally, without having tested it myself, in conventional modes (single point AF/expanded point AF) I would probably rank the R5 about the same maybe a bit better than my 5DIV in good light. I agree that I have noticed a fall off with the R5 in moderate to low light. Not bad, but I think there may be something there similar to what you have observed. Whether the PDAF falls off as much or at the same levels of brightness, I am not sure, but it is well established PDAF also falls off with decreasing light (Popular photo used to run a great test on new camera bodies showing this). But, then with my personal experience, I know the 5DIV falls off in dark settings and the R5 has been doing pretty well there. So, it might take some series testing to really figure out which is better when and were. As for shallow DOF, I have not had much of an issue when trying to use my f/1.4 lenses, but those situations are much more limited as I've primarily been using my R5 for birds. In fact, I have yet to try human Eye-AF (pandemic times). But animal eye AF has worked well on birds, dogs, cats, and a fox so far.

Let me know if there is some sort of test you'd be interested in. I do not have the 1DXIII (it was a loaner), but I could try the 5DIV vs the R5.

Also, let's not lose track of the fact that the PDAF system in the 1DXIII is amazing. I still get excited thinking of how quickly it came into focus.