Hepatocellular carcinoma (HCC) is the most common primary liver cancer and remains difficult to treat in advanced stages because of tumor heterogeneity and an immunosuppressive microenvironment. Chimeric antigen receptor (CAR) T-cell therapy is an emerging strategy for solid tumors, but optimal CAR architecture for HCC remains under investigation. This study compared second-generation (anti-GPC3 CAR2), third-generation (anti-GPC3 CAR3), and fourth-generation (anti-GPC3 CAR4) CAR T-cells targeting glypican-3 (GPC3). Immunohistochemistry of liver biopsy specimens from 195 HCC patients showed that 82.6% were GPC3-positive. Anti-GPC3 CAR4 T-cells showed the strongest cytolytic activity against GPC3-positive target cells in both 2D monolayer and 3D spheroid models. At an effector-to-target ratio of 10:1, anti-GPC3 CAR4 showed significantly higher cytolytic activity against HepG2 cells (52.55 ± 10.04%, p = 0.003) and HeLa-GPC3 cells (64.45 ± 7.81%, p = 0.0002) than against GPC3-negative controls. In 3D spheroids, anti-GPC3 CAR4 also showed greater activity against GPC3-positive HepG2 (4.54 ± 0.48, p < 0.05) and HeLa-GPC3 spheroids (4.823 ± 0.18, p <  0.001). Anti-GPC3 CAR4 T-cells retained antigen-dependent proliferation with a moderated acute cytokine profile. After acute antigen exposure, anti-GPC3 CAR4 T-cells showed lower PD-1 and higher TIM-3 than anti-GPC3 CAR2 T-cells. Collectively, anti-GPC3 CAR4 T-cells were the most active construct in vitro ; in vivo durability, biodistribution, and safety require validation.