Unless you want to go through different drives and testing them. Reviews are one thing but, real world use is another completely. Testing a baren drive and one that's 50-75% full is a different story as well. Moving actual files vs synthetic tests is going to behave differently as well.
#'s are only part of the game. Some drives bench exceptionally but fail miserably when you put data on them.
so, here I tested 2 different SN770 drives w/ and w/o data on the drive and they're on par with your 4K writes in a Gen3 slot. If I moved the SN770 over to a Gen4 sot they might blow away the readings comparatively with more bandwidth available. I just didn't feel like doing it as my focus was comparing the speed between the internal slot and the TB4 enclosure. They're both a tight grouping for speeds which satified my need for testing the drive + enclosure to make sure it's as quick as it can be.
I also tested Phison based drives and others in the enclosure and they had 30-50% deficits in speeds whether from the controller / dram / chips.
I went through a similar testing with internal drives on my server with another Phison based drive before swapping it out with the SN850 but, the other issue was the testing program on the server couldn't keep up with either drive so I ended up using a different method for testing and got more realistic top end numbers that weren't bottlenecked.
When it comes to the price spread on 1TB drives low being the SN770 @ $80 and the higher end drives closer to $200 it comes down to use, If the SN770 was an option back when I built either system I would have tried it and compared things and probably saved ~$100 or more on drives.