This chapter introduces the reader to a procedural strategy that should be followed to approach the model selection phase in an effective and organised manner. The starting point is a clear understanding of the task we want to perform and how integrating LLMs with the search solution should benefit the system. After an initial selection based on task-solving capabilities, the set of candidates must be filtered by the type of license to guarantee legal compatibility with the intended usage. Finally, the training data should be evaluated both in terms of form, domain and language, to be as close as possible to the domain of interest. The chapter closes by exploring useful platforms to compare LLMs side by side and help users and developers with such a daunting decision that’s key to the success of a project.