The effort and expertise required for manually crafting the models for model-based testing (MBT) is a major obstacle slowing down its industrial adoption. For implemented and executable systems, there are approaches to automate some part or even the whole process of creating the models for MBT. Recently, using extracted models for testing graphical user interface (GUI) applications has been a popular area of research, but most of the proposed approaches have limitations and restrictions on what can be modeled, and the software industry has not adopted these approaches. In this paper, we try to identify the gaps between the academic approaches and tools and industrial requirements hindering the industrial adoption, and try to suggest practical solutions to the identified gaps.