Infant habituation—the decrease in attention to repeated stimuli—provides key insights into early cognitive development and individual differences in attention, learning, and memory. While previous studies have made important contributions using traditional methods, new modeling approaches allow for more precise estimation of individual variability in habituation patterns. This study presents a proof-of-principle implementation of multilevel modeling combined with non-linear curve models—specifically the Thomas & Gilmore and Dannemiller functions—to estimate individual-specific habituation parameters. Using the ManyBabies 3 dataset, we validate these models through simulation, parameter recovery, and model comparison. The approach enables detailed investigation of habituation speed, initial attention, and asymptotic looking time, and explores age-related patterns and responder subgroups to highlight the potential of this framework.