Frequent forced switching between tasks has been shown to reduce switch costs and increase voluntary switch rates. So far, however, the boundary conditions of the influence of forced task switching on voluntary task switching are unknown. Thus, the present study was aimed to test different aspects of generalizability (across items, tasks, and time) of switching-induced flexibility established in forced-choice trials on voluntary switching in free-choice trials. To this end, stimuli and tasks were systematically varied between forced- and free-choice trials in a hybrid task-switching paradigm. In a series of three experiments, we manipulated forced-choice switch probability (25% vs. 75%) and found that switching-induced flexibility generalizes to new items, but arguably not to new tasks. This task-specific effect is rather short-lived, limited to the first free-choice trial following a forced-choice trial. Underlying mechanisms of switching-induced flexibility, the versatility of flexibility and implications for the trainability of cognitive flexibility are discussed.