UGBW is difficult to test well on ATE, even
more so at probe. Often this would be a
GNT param, and part acceptance would
use a correlated proxy such as slew rate.
I wonder whether mass market consumer op
amps see packaged test at all, anymore.
Many sub-$/part products that run big volume
don't even probe every die and if yield on what
-is- probed is below par, they just scrap the
wafer before any more money goes into it.
This might be a fly in the "bin it out for higher
price" ointment, even if the part is capable.