This paper documents a dimension of crosslinguistic variation among Universal Free Choice Items. In English, the distribution and interpretation of *any* DPs containing a numeral (‘numeral *any*’) differs from that of *any* DPs with no numeral (Dayal 2005, 2013; Chierchia 2013). In contrast, the Farsi counterparts of *any* and numeral *any* mirror each other. Two competing analyses of the contrast between *any* and numeral *any*—the Wide Scope Constraint Analysis (Chierchia 2013) and the Viability Constraint Analysis (Dayal 2013)—are assessed against the Farsi data. The paper shows that, with minimal extensions, either analysis can capture the behavior of the Farsi counterpart of numeral *any* with distributive predicates. The situation changes, however, when the minimally modified analyses are assessed with respect to sentences with collective predicates: the minimally extended Wide Scope Constraint Analysis captures the attested interpretation of those sentences, but the Viability Constraint Analysis rules them out, undergenerating.