Energy flexibility in buildings has been increasingly viewed as a key solution to address the power imbalance between supply and demand sides in electric grids, which is caused by the growing share of intermittent renewable energy generation. This chapter investigates the energy flexibility of buildings at individual and cluster levels. At the individual building level, we examine the energy flexibility of two types of space cooling and heating systems in residential buildings, including variable-speed heat pumps in Hong Kong and floor heating systems in Denmark, by using the advanced model predictive control technique. The proposed model predictive controller helps to improve thermal comfort at the beginning of occupancy, reduce power consumption during peak demand periods, and reduce daily electricity costs. We then introduce a novel supply-based feedback control to enable commercial buildings to provide fast power reduction in response to short-term or even immediate requests from smart grids. At the building cluster level, we review the state-of-the-art classification, architectures, and techniques of neighborhood-level coordination and negotiation in residential microgrids. Moreover, we explain the necessity and a method to realize the cluster-level building demand management in a distributed mode using game theory-based approaches.