Penicillin G acylase (PGA) is a key biocatalyst for the enzymatic production of β-lactam antibiotics, which can not only catalyze the synthesis of β-lactam antibiotics but also catalyze the hydrolysis of the products to prepare semi-synthetic antibiotic intermediates. However, the high hydrolysis and low synthesis activities of natural PGAs severely hinder their industrial application. In this study, a combinatorial directed evolution strategy was employed to obtain new PGAs with outstanding performances. The best mutant βF24G/βW154G was obtained from the PGA of Achromobacter sp., which exhibited approximately a 129.62-fold and a 52.55-fold increase in specific activity and synthesis/hydrolysis ratio, respectively, compared to the wild-type AsPGA. Thereafter, this mutant was used to synthesize amoxicillin, cefadroxil, and ampicillin; all conversions > 99% were accomplished in 90–135 min with almost no secondary hydrolysis byproducts produced in the reaction. Molecular dynamics simulation and substrate pocket calculation revealed that substitution of the smallest glycine residue at βF24 and βW154 expanded the binding pocket, thereby facilitating the entry and release of substrates and products. Therefore, this novel mutant is a promising catalyst for the large-scale production of β-lactam antibiotics.