They compared 6800XT to consoles in photo mode benchmarking coverage, but a) 6800XT still achieved the same/slightly better worst case ("Corridor of Doom") performance than consoles, b) it did so while running RT at twice the resolution (consoles use checkerboarded RT, PC uses full res) + higher LOD level (as consoles run below PC's low LOD) + an older engine build + using regular DX12 instead of specialized console APIs. So no, XSX didn't run better than 6800XT, 6800XT just ran "less better" than expected. And TBH these results aren't even directly comparable due to differences in settings and software, which DF mentioned clearly in their footage, but even with those higher settings and worse software stack it never ran worse. Which is underwhelming, as they said, but not worse.
BTW I just beat Control on a slightly underclocked 6900XT (which should result in <10% better performance than 6800XT) at high details + RT reflections, RT transparent reflections and RT debris and got a mostly stable 1440p60 with dips to 50s and sometimes momentary 40s in intense combat. This would be up to 50% better than what consoles offer, even though both rasterization and RT quality were higher. Obviously with unlocked framerate consoles would do better, as photo mode suggests, so the real advantage is probably closer to maybe 20-30% (I got 70-80fps where consoles did 50-60) over XSX most of the time, but it is still there. I suspect I could get another 10-20% if exact console settings were available on PC.
This footage can be found here, the link should point exactly to the beginning of PC/console comparison: