A comparison was made between $Gaia$ magnitudes and magnitudes obtained from ground-based observations for astrometric radio sources . The comparison showed that these magnitudes often not agree well. There may be several reasons for this disagreement. Nevertheless, such an analysis can serve as an additional filter for verification of the object cross-identification. On the other hand, it can help to detect possible errors in optical magnitudes of astrometric radio sources coming from unreliable or inconsistent data sources.