The K3 has exceptionally clean keying sidebands, among the best if not the best from what I read, but there was much more to the ARRL tx testing. I would also consider the other test results.
For another perspective, here’s a link to Adam Farson/AB4OJ’s test report site, which contains his Flex 6700 test report (note that tx testing was conducted at 100W):
BTW, Adam also has a new presentation on the site (under Test Methods and Procedures): A New Look at SDR [receiver] Testing
I just did a quick search here and found this very interesting discussion:
Gerald, thank you once more for your contribute. Pls evaluate to leave the adjustments as they are and instead modify the technical specifications in the way they are written.