The -30dbc for IMD of third order is good. Whats no so good is only 2W.
It was actually hard to find that data for a lot of the QRP/low power radios.
For many it doesn't exist.
I never got as far as to measure that for uBitx. I did it for the old bitx20
(2006 version) and it was not at all good at 4W and -20dbc. Reworking
the amp for better stability as that design oscillated when bias hit 60ma
improved matters to -25dbc at 5W.
A lot of the newer 100W radios out there are barely -26dbc.
Nominally its measured at max stated output power or 10W for ubitx,
and anything better than -26dbc is as good as a lot of radios.
I suspect on 20M and up that will be limited by available power without
hitting 1DB compression point. Once you hit compression all bets are off.