With the growing size and complexity of systems under design, industry needs a generation of Model-Driven Engineering (MDE) tools, especially model query and transformation, with the proven capability to handle large-scale scenarios. While researchers are proposing several technical solutions in this sense, the community lacks a set of shared scalability benchmarks, that would simplify quantitative assessment of advancements and enable cross-evaluation of different proposals. Benchmarks in previous work have been synthesized to stress specific features of model management, lacking both generality and industrial validity. In this paper, we initiate an effort to define a set of shared benchmarks, gathering queries and transformations from real-world MDE case studies. We make these case available to community evaluation via a public MDE benchmark repository.