In order to most effectively use models at runtime, self-aware computing systems need increasingly powerful ways of observing their operational environment and their own performance and behavior and then building and refining their own models accordingly. An inherent principle of self-aware computing systems is having diverse feedback loops, which build a causal connection between the computing system and a reflective layer. The computing system is continously observed and, based on this, the system is able to update and modify its models to reason about its goals, context, operational environment and its own resources, decisions and actions. To effectively and efficiently realize these feedback loops, models and especially modifiable and updatable models@runtime are essential. The models@run. time paradigm proposes to use runtime models as abstractions of the computing system for the purpose of more efficient reasoning upon both its runtime observations and learned knowledge. Hence, models@runtime is especially looking for more innovative approaches to the causal connection between the system and the runtime model, with particular focus on a transaction concept for this causal connection for such issues as timing, roll-back ability and data-consistency. The goal of this workshop is to provide a bridging podium for researchers working in the area of self-awareness, self-modelling, autonomous and organic computing, as well as self-adaptive and self-organizing systems with a focus on runtime representations that can be used by the system to reason about its goals, context, operational environment and its own resources, decisions and actions.