Multiscale methods have enjoyed a well-celebrated place in the computational chemists' toolbox, while the next generation of so-called adaptive quantum mechanics/molecular mechanics (QM/MM) methods were being developed on the fringe for the past 20 years. Adaptive QM/MM methods hold the promise of extending the range of applicability, currently at the expense of complexity and computational scaling. There are a number of challenges in the area; firstly, the ability to partition a system on the fly has led to the distance, number, density, and stress-based approaches. Secondly, methods that smoothen the transition from a QM to an MM region using a single configuration include Hot-Spot, ONIOM-XS, and time-adaptive. Alternatively, more computationally expensive methods that smoothen the transition region based on multiple configurations are the permuted and sorted adaptive permutation, difference-based adaptive, and size-consistent multipartitioning methods. There exist three alternative methods that avoid issues of smoothening altogether; they are the so-called buffered force, ABRUPT, and the flexible inner restraint methods. In this advanced review, an overview on the available methods, a number of applications, and the remaining challenges are discussed.