The grid-connected quasi-single-stage converter (QSSC) provides a direct power flow path from low-voltage (LV) energy storage systems (ESS) to ac–dc converters, resulting in reduced power conversion stage for high-efficiency ESS integration. However, due to the dc-link voltage ratio between the two ports varies with the ESS state of charge, the voltage vectors in the ac side are unevenly distributed in the QSSC. This nonlinearity brings some challenges in control and modulation design for the ac–dc part of QSSC. In this article, a virtual two-level converter model considering the two ports as a single dc source is proposed, and model predictive power control (MPPC) based on the virtual two-level model is proposed to enhance the grid current performance. The duty cycle of each phase is calculated based on the active and reactive power tracking error minimization in the two-level converter frame, and the duty cycle for each switching pair is obtained by minimizing the power processed by the dc–dc converter. In this way, the complicated implementation of MPPC for a three-level converter is avoided, and the controller design of the QSSC can be simplified. The effectiveness of the proposed virtual two-level converter model and MPPC method are verified with experimental results.