At this very moment, I’d say the 4Kxx versions, mostly because the quality across the whole feature is basically the same throughout, whereas Despecialized varies from 1080p to barely DVD quality. Having said that, if Harmy can pull off what he’s trying to do with Star Wars Despecialized 3.0 and beyond, his versions would become my preference, especially since I’m still watching stuff at 1080p so any minor differences between the sources at 4K would become negligible.
I wouldn’t expect too much improvement in terms of “quality seams” in newer Despecialized releases. Most of the original footage in the current releases of Harmy’s trilogy is from 35mm print scans (not necessarily from the 4Kxx project, but comparable amounts of detail). In Jedi, for example, there’s only one original element taken from the GOUT, and he only did that because the 35mm theatrical scans didn’t have appreciably more detail than the GOUT for that shot (think about THAT for a second!). The optical duplication process used to create theatrical prints from 35mm negatives simply blurs out fine detail, and it doesn’t matter what resolution you scan a projection print at, you’re not going to get that detail back. When you notice certain parts of Despecialized don’t even seem to have 720p worth of fine detail, more often than not, that’s from a 1080p or 4K 35mm print scan, not the GOUT.
That said, since the UHD’s aren’t covered with fake grain like the Blu-rays, it may be possible to grain-match the various sources without over-graining the 35mm footage, which might help to a small degree. Also, the Jedi print used by the 4K83 project is a bit of a unicorn, so there may be a hair’s improvement to be gained from using 4K83 as a source. But these will be very, very incremental improvements at best. The huge quality jump will come from the UHD source, and that will make the quality seams harder to hide, not easier. Not that Harmy may not have learned a trick or two in the intervening years.