Event cognition theories suggest that the structure of events is hierarchically organized. While working event models process sensory (modal) information, more abstract (amodal) information is processed at higher-level event models in long-term memory. Studies also show that event descriptions become more abstract when the grain of the event shifts from fine context (e.g., “breakfast”) to coarse context (e.g., “morning routine”). However, it is an open question when (and how) this abstraction occurs and when event models become amodal. Our preregistered experiment tested the influence of context grain and presentation modality on the nature of event models. We presented participants (N = 88) with fine or coarse events in either text or video format. Their task was to decide, as fast as possible, if the following video stimulus (target or distractor) fits the event presented before. Reaction times to hits were used as the dependent variable. Results showed a main effect of context grain. Participants responded faster to fine events than to coarse events. The main effect of modality was also significant, with participants responding faster when the target modality was video (compared to text). Further, we observed no interaction. These results suggest that (1) the fine context triggers modal representations of events, (2) the coarse context (as realized in the present experiment) does not lead to more amodal representations, and (3) that viewers benefit from a match in context modality. We discuss the results in the context of present event cognition theories.