dpb writes:

Heat loss is heat loss

That's naive. Performance depends on the design, which varies for cooling
room air, versus wall conduction losses, making ice, defrosting, etc.
Efficiency has more to do with those parameters than any basic heat pump
efficiency. That model A is better than B for the few modes tested by the
DOE, does not mean that A beats B for other modes.

Indeed, the opposite is quite to be expected, since the design will be
optimized to the DOE fantasy test, which appears on a big yellow immunized
sticker, rather than performance under real conditions, which most
consumers never measure. You know, putting stuff inside, making ice,
opening the door. The DOE test forces designs that idle cheaply, rather
than ones that cheaply recover from intrusions, defrost, or chill or freeze
contents.

Quite typically the ultra-efficient designs get the last bit of efficiency
from complex mechanisms that are the first to fail and fall-back, leaving
you worse off.