your supplier is right, perfomance in the lab will be better with 7602
7602/7612 have hardware (lets think it's a hardware till it's not fixed by updating microcode) issues with interference, and this weakness multiplied by missing calibration data for your board (7612 have the same issues but 5GHz interference is much more rare than 2.4GHz)
7603 is better cuz it's much more stable in case of interferences (other 2G AP's near, microwave ovens, radio nannies, personal radio stations, LTE base stations near, SATA3/USB3 interferences etc) and it's much more stable without proper calibration (it's much more CAL FREE than 7602)
so in the REAL LIFE mt7603 will give you much less troubles with wifi stability
now im testing 220.127.116.11 drivers tweaked by Eugeny Manachkin (sfstudio/wive-ng-mt) with Andy Padavan's help, they modified some noise/interference logic and first feelings the radio become more stable, but still not ideal and they say it can't be resolved by driver only modifications.