High-tech embedded systems increasingly depend on software: software controls, connects, and monitors al- most every aspect of system operation. Therefore, overall system quality is more and more determined by software quality, and advances in software testing directly improve the whole development process as well as the quality of the developed products. Model-based testing, where tests are automatically generated from a model of the system under test, is a very promising approach to detect more bugs faster and cheaper. Together with the partners ASML, Océ, PANalytical, and TNO-ESI, SUMBAT will extend the application of model-based testing to large and complex, high-tech embedded software systems, involving millions of lines of code, distribu- tion, concurrency, uncertainty, and complex interfaces. More specifically, SUMBAT will integrate model-based testing with automata learning to extend partial models, provide test generation from models that combine millions of states with complex and large data structures, and integrate user-profile based test selection into model-based testing so that test cases are generated that reflect the real usage of the system.