Hepatocellular carcinoma (HCC) is often detected at advanced stages among patients with hepatitis B virus (HBV), underscoring the urgency for more precise surveillance tests. Here, we compare the clinical performance of the novel - GAAD (gender [biological sex], age, alpha-fetoprotein [AFP], protein-induced by vitamin K absence-II [PIVKA-II]) and GALAD (gender [biological sex], age, AFP, Lens-culinaris AFP [AFP-L3]), PIVKA-II) algorithms to assess the utility of AFP-L3 for distinguishing HCC from benign chronic liver disease (CLD) in Chinese patients with predominantly chronic HBV infection. Eligible adults were enrolled, and biomarkers were measured using Elecsys (Cobas) or µTASWAKO assays. In total, 411 participants provided serum samples (HCC, n = 176 [early-stage, n = 110]; CLD, n = 136; specificity n = 101). HBV was the underlying disease etiology for most participants (HCC, 95%; benign CLD, 72%). For GAAD (Cobas), GALAD (Cobas), and GALAD (µTASWAKO), AUCs were 93.1% (95% CI: 90.0–96.2), 93.2% (90.0–96.3), and 92.7% (88.4–96.9) for early-stage, and 95.6% (93.6–97.6), 95.6% (93.6–97.7), and 95.8% (93.2–98.3) for all-stage HCC, versus CLD, respectively. Interestingly, both GAAD and GALAD algorithms demonstrated comparable diagnostic performance regardless of disease etiology (HBV vs. non-HBV), presence of cirrhosis, geographic region, and within pan-tumor specificity panels (p < 0.001), indicating AFP-L3 may have a negligible role in HCC surveillance.