Millimeter-wave (mmWave) automotive radar can detect the surrounding environment using highly directional, high-frequency electromagnetic waves. With these advantages, mmWave radar has become an important component of autonomous driving and integrated sensing systems. Designing systems and sensing algorithms requires the application of realistic channel models and simulation. In this paper, the propagation channel is measured, characterized, and modeled for mmWave automotive radar with typical configurations in scenarios. Based on the channel measurements in urban street and expressway environments, ray-tracing (RT) technology is verified for modeling important objects regarding radar cross-section and echo power. A hybrid channel model is proposed by integrating RT with the stochastic modeling of targets and surrounding environments, which significantly improves simulation efficiency and provides more flexibility for virtual tests in various complex environments.