As automated driving and connected-vehicle technologies proliferate, mixed traffic has become the norm, yet traditional homogeneity-based dynamic traffic assignment fails to capture capacity fluctuations, congestion propagation, and nonlinear phase transitions induced by key parameter shifts. This paper reviews advances in modeling, equilibrium, evolution, and solution methods for mixed traffic, and clarifies the mechanisms, properties, and scope for regulation of the user equilibrium–system optimum continuum. At the modeling level, we compare heterogeneity across micro, meso, and macro scales and introduce a decoupled “automation × connectivity” framework. In equilibrium, we synthesize three complementary formulations: game-theoretic, variational-inequality, and multi-objective. For solution, we compare analytical, simulation, and learning approaches. We establish existence, stability, attainability, and other related properties of mixed-traffic equilibria at within-day and day-to-day scales, identify stage-wise critical thresholds, and surface three limitations: reliance on binary partition models, efficiency-centric objectives, and shallow treatment of evolution–control mechanisms. Finally, we recommend a two-dimensional heterogeneity framework, a multi-objective dynamic equilibrium, and coordinated open- plus closed-loop control to strengthen both mechanistic interpretability and real-world adaptability of dynamic traffic assignment.