Contemporary epistemology typically relies on idealizing conditions that are overly stringent, making it difficult (actually, impossible) for real subjects to meet its normative demands. It raises the issue that idealizing epistemic norms aren’t genuinely normatively binding, given that “ought” implies “can”. Advocates of bounded rationality, on the other hand, seem incapable of offering a “robust” normative account of epistemic evaluation that do not slip into full pragmatism. In this talk, I defend a reformist account of epistemic rationality that is both purely epistemic and resource-bound. I argue that (i) any state of epistemic rationality must be evaluated in terms of the processes of rational inquiry that lead to it, and (ii) for resource-bound subjects, some cognitive costs are essential to rational inquiry, and hence are constitutive of the evaluation of bounded subjects’ epistemic rationality, thereby giving rise to a robust epistemic norm that is non-pragmatist and yet cost-sensitive.