This chapter of the book discusses potential artificial-intelligence-based enhancements for the Physical Layer (PHY) of 6G systems. Given a dataset of wireless channels representing the radio propagation environment, generative models can be utilized to capture the underlying probability density function of the wireless channels across the entire base station's cell propagation environment. This knowledge can then be used to enhance various PHY applications. While many generative modeling techniques have gained popularity in recent years within wireless communications, this chapter focuses on Gaussian Mixture Models (GMMs). A significant advantage of GMMs is its analytic representation, which allows straightforward integration into system models describing different PHY applications. Moreover, GMMs can easily adapt to various system parameters, such as signal-to-noise ratio, or the number of feedback bits, pilots, and mobile terminals. Leveraging model-based insights, GMMs can be tailored for different applications, minimizing memory and model-transfer overhead. In this chapter, we present GMM-based enhancements for four applications: channel estimation, channel prediction, precoder design, and pilot design.