Modelling is undoubtedly a rewarding activity in Software Engineering. The models vary in notation, expressiveness and level of abstraction. Models can be used for a variety of purposes, such as abstract thinking, code generation and test case generation.
In model-based testing (MBT), models function as oracles, allowing you to decide whether test cases pass or fail. Models are usually created on purpose for MBT, which means that it is uncommon to reuse models that support development activities. This may be one of the reasons why it may be difficult to adopt MBT in the industry. It may not seem clear whether the test modelling effort will yield a return on investment.
This talk describes a model-based testing research path that starts using textual models, then graphical, and finally, pattern-based models. Along the way, the expressiveness power and the modelling effort diminish, while the purpose of the test becomes more focused, but applicable to a wide range of software applications.
So yes, there is no doubt that software engineers need models. However, the nature and purpose of the models must agree, and the relationship between the modelling effort and the return on investment must be balanced.