Socio-dynamic situations require a balance between response execution and action inhibition. Nonadaptive imbalance between response inhibition and execution exists in neurodevelopmental and neuropsychological disorders. To investigate the underlying neural mechanisms of cognitive control and the related deficits, comparative studies in human and nonhuman primates are crucial. Previous stop-signal tasks in humans and macaque monkeys have examined response execution (response time (RT) and accuracy in Go trials) and action inhibition (stop-signal reaction time (SSRT)). Even though marmosets are generally considered suitable translational animal models for research on social and cognitive deficits, their ability to inhibit behavior remains poorly characterized. We developed a marmoset stop-signal task, in which RT could be measured at millisecond resolution. All four marmosets performed many repeated Go trials with high accuracy (≥ 70%). Additionally, all marmosets successfully performed Stop trials. Using a performance-dependent tracking procedure, the accuracy in Stop trials was maintained around 50%, which enabled reliable SSRT estimates in marmosets for the first time. The mean SSRT values across sessions ranged between 677 and 1464 ms across the four marmosets. We also validated the suitability and practicality of this novel task for examining executive functions by testing the effects of a natural hormone, oxytocin, on response execution and action inhibition in marmosets. This marmoset model, for reliable (millisecond resolution) assessment of the balance between response execution and inhibition, will further facilitate studying the developmental alterations in inhibition ability and examining the effects of various contextual and environmental factors on marmosets' executive functions.