The SU(4) x O(4) and the ``flipped`` SU(5) x U(1) models seem to be the only possible Grand Universal Theories (GUT`s) derivable from string theories with Kac-Moody level K=1. Naively, the SU(4) x O(4) model, at least in its minimal GUT version, is characterized by the lack of any mixing in the quark sector. In this ``Comment`` we show that, although some mixing may be generated as a consequence of large vacuum-expectation-values for the scalar partners of the right-handed neutrinos, it turns out to be too small by several orders of magnitude, in net contrast with our experimental information concerning the Cabibbo mixing. Our result, which therefore rules out the minimal SU(4) x O(4) GUT model, also applies to ``flipped`` SU(5) x U(1) in the case of the embedding in SO(10). (Author).