This all depends on expectations, construction and
what limits reliability under those conditions. I have
personally run integrated JFETs and MOSFETs at
300C for very long periods of time and everything
worked just peachy (no doubt in part because the
process technologies were developed for HiRel
applications). But power devices are another thing
entirely, with internal temp rise on top of case
temp and the potential for some designed-to-the-bone
feature such as body access resistance being just
barely OK at 175C, and avalanche burnout at just
a hair more temp*voltage, perhaps, when Rbx goes
higher along with parasitic BJT leakage and gain,
while the Vbe (Vbs) needed to initiate snapback
goes down.
Specs tell a story. They leave out the pictures
sometimes.