Gender gaps in Science, Technology, Engineering and Mathematics (STEM) emerge as early as elementary school, highlighting the need for early interventions. Recent work suggests that a pretend play intervention, in which girls roleplayed as a hardworking scientist increased their persistence in a novel science activity. However, instead of highlighting scientists' hard work, the pervasive cultural narrative portrays scientists as brilliant, a trait that girls associate less with their own gender than boys. The present study investigates whether pretending to be a brilliant scientist can also boost girls' persistence. Experiment 1 tested a large and diverse sample of four- to seven-year-olds (N = 325, 164 girls, 55% White). Children played a science game in one of three conditions: as themselves (no-roleplay condition), as a hardworking scientist (dedication condition), or as a brilliant scientist (brilliance condition). Results showed that children in both the dedication and brilliance conditions persisted longer in the science activity than those in the no-roleplay condition. This effect was mainly driven by girls. Thus, pretend play of science role models enhances children's science engagement, regardless of the role models' characteristics. Experiment 2 (N = 160, 82 girls, 50% White) revealed that roleplaying as an artist did not yield the same effect, suggesting that pretending to be scientists, not pretend play in general, increased children's persistence in science activities. These findings have broad implications for ways to mitigate the gender gap in science. We discuss possible mechanisms driving the role of pretend play in boosting children's science engagement.