... The antennas on them are trash..
Normally I'd agree, although as you mentioned height and clearing obstructions are more important at UHF than raw power between 2-5 watts.
I think manufacturing quality is also important, especially when it comes to the receiver.
I just ran an interesting test. Originally I picked up a Midland GXT1000 and like it so much that I also got the smaller LXT490. I also have a Radio Shack Pro-106 digital scanner, and a Uniden 396XT digital scanner.
For the test, I listened to a very weak GMRS repeater out in the boonies, that barely popped open the squelch on the smaller 490, and ran to my other radios for testing and here is how they came out from best to worst:
LXT490
GXT1000
Uniden 396XT scanner with Comet HT-55 dualband 155/465 mhz
Radio Shack Pro-106 with Diamond RH-77CA (dualband amateur 2m/440)
The 490 recovered the audio for 100% copy, although weakly
The 1000 recovered the audio, but had to run with open squelch and find a hot spot
The 396XT knew that something was there, and even broke the minimum squelch, but no recoverable audio was heard.
RS Pro-106. Quiet - nothing heard even with open squelch and no hot spots to be found unless I went up on the roof.
Each radio is being powered by capacity-matched Sanyo Eneloops.
What blows my mind is that a $20 bubblepack actually had a better receiver than a $500 digital scanner for frs/gmrs, running with highly popular tx-capable antennas.
The 490 has a slightly smaller antenna than the 1000, yet it performed better (at least on receive - suspect a higher sensitivity / better filtered receiver on the 490.
What this showed to me was that the quality of the receiver at least, has much to do with it as well. I'm sure if I cut open the 490/1000 antenna, I'd be shocked, but at least from an rx-only standpoint, the best performing radio was the one with the smallest antenna for this weak-signal test.