When we apply scientific theories to the world in description, prediction, and explanation, we often use models that are appropriate to the phenomena in rather different ways. In some cases it would even appear that models of (ostensibly) one and the same thing are incompatible with one another. This represents a prima facie challenge to scientific realism, the view that our best scientific characterizations of the world are true or approximately true. In recent years a number of authors have attempted to square realism with the use of incompatible models by suggesting that they should simply be regarded as furnishing different perspectives on the relevant phenomena. In this talk, Chakravartty will argue that while a general impulse toward pluralism about scientific knowledge is well motivated here, perspectivism is not. Chakravartty considers some examples of what realists should say instead, in different sorts of cases.