Model-Driven Engineering (MDE) is an approach to software engineering in which the model is considered a central concept. Its main objective is to generate all or part of an application’s code from models. Up to now, MDE is an active research topic that makes intensive use of metamodels, models, and model transformations. In this context, the generation of models is generally done manually to ensure conformity with their metamodels. Several works have been proposed to automate this process, but none of them can ensure complete automation without starting from a set of models already defined by the user or with a good verification of all conformity constraints. Automated generation of metamodel instances significantly reduces the time and effort required compared to manual creation, and the generated models offer several advantages in MDE and domain-specific language development, such as increasing productivity and improving model consistency. In this work, our goal is not only to automatically generate the models and verify all conformity constraints, but also to explore most machine learning techniques to solve modeling problems in the MDE context.