CONTEXT: Visual modeling languages like UML and BPMN are often used to represent and communicate knowledge about a software systems and business processes. In this function, modeling languages compete with prose texts expressed in natural languages. This begs the question, which of the two is "better". We argue that one particularly important feature of natural languages is the epistemic mood, i.e., the modality expressing degrees of reality: natural languages allow to express "is" as well as "could be" and "will be". Modeling languages like the UML offer no such flexibility. AIM: We strive to improve modeling utility and practice by introducing modalities in modeling languages. We show how this can be done with little effort for existing modeling languages and tools, and demonstrate how this novel facility can be exploited for practical purposes. METHOD: We define the notion model modality and explore practical modeling examples to arrive at a set of common (useful) modalities. We design visualizations for modalities to maximize understandability and minimize effort. We study how they can be integrated into existing modeling languages like UML or BPMN. RESULTS: We find several obvious model modalities with straightforward practical applications. They afford natural visualizations and can be applied in existing tools. CONCLUSIONS: Contemporary modeling can be enriched by modality annotations resulting in increased expressiveness of models at minimal extra effort. Our approach integrates with existing methods, notations, and tools.