I would wait for someone like der8auer to do a delid on these things to see what's under the surface. The 10c dice are a new die, while the 8c and 6c dice look like the old 8c Coffee Lake die all over again. Not sure that the 10700k will get you anything that a 9900k won't except a higher TDP (so you can set it to strict 125W in UEFI and get higher clocks than what you would get with a 9900k set to 95W TDP).
Or you could go by the ordering and spec information listed on ARK. Just
as the leaks had indicated, the entire Comet Lake-S product stack is based on two dies: the 10-core stepping Q0 and the 6-core stepping G1. They are not using any previous generation Coffee Lake dies or the recently released CML-H 8+2 R1 stepping at this time.
Core i5-10600KF and above are Q0 stepping and likely thinned die with solder TIM.
Core i5-10600 and below are G1 stepping and probably conventional die with paste TIM (which is fine seeing as they are all locked SKUs with 6 or fewer cores anyway).
Core i5-10400 and Core i5-10400F are the only SKUs that currently list spec numbers for both Q0 and G1 steppings.
What I'm still a little unclear on is why early benchmarks appear to show so many Comet Lake HP die steppings. The old
Tweakers roadmap leak and more recent
production dashboard leak led me to believe that Comet Lake-S would be a Q1'20 launch and Comet Lake-H would follow in Q2'20. It also looked like CML-H would be based on the same (or at least very similar) 10+2 HP die as CML-S. Once it became clear that CML-H would launch before the S Series, and that it would use a new R1 stepping of the previous R0 Coffee Lake Refresh 8+2 die, I figured this was entirely due to Intel needing to defend a critical market segment against the onslaught of AMD's Renoir.
However, despite ARK only listing Q0 and G1 steppings for production CML-S SKUs, benchmarks show entries for "Family 6 Model 165" steppings 0, 1, 3, 4, and 5. So were the delays actually caused by Intel needing to do multiple respins of the CML-S dies way late in the cycle in order to fix hardware bugs? Was the R1 stepping a plan B when they realized the 10+2 HP die had problems and they'd have nothing new ready in time to counter the Renoir launch?