To clarify the effect of individual differences of fluid intelligence (Gf) on attention resource allocation, the present study proposed a new hypothesis (i.e., integrated control hypothesis) based on previous studies and provided preliminary empirical evidence through a pupillometry study. Results showed that task types and task difficulty played distinct roles in that relationship. In the exploitation task, higher Gf individuals allocated significantly fewer attention resources than those with lower Gf in all three difficulty levels. In contrast, in the exploration task, smarter people allocated significantly fewer resources in low-difficulty trials whereas more in medium- and high-difficulty trials, with this phenomenon more significant in male subjects. In conclusion, this study suggested that high Gf individuals tend to control their attention state in tasks with diverse demands, making them dynamically optimize the use of attention resources and flexibly adapt to changing conditions.