Duchenne muscular dystrophy (DMD) is a fatal, X-linked genetic disease due to mutations in the DMD gene and has become the most common neuromuscular disorder in childhood with a prevalence reaching 1 in ~3500 male newborns worldwide. However, there is no curative therapy for this devastating disease. Among thousands of DMD-causing mutations, the skipping and reframing of exon 51 is estimated to therapeutically benefit at least 13-14% of all DMD cases worldwide. In this study, we generated a humanized DMD mouse model by replacing mouse exon 51 with human exon 51 and deleting mouse exon 52, thereby disrupting the open reading frame of the dystrophin protein. The DMD mouse model exhibited phenotypes highly similar to those of DMD patients, suggesting this model was suitable for the analysis of potential intervention strategies. We subsequently showed that high-fidelity (hf)Cas12Max-mediated single-cut genome editing efficiently targeted the splice donor (SD) or splice acceptor (SA) sites of human DMD exon 51 in-vitro. To demonstrate the in-vivo single-cut editing efficiency of hfCas12Max, we used an all-in-one adeno-associated viral (AAV) vector for delivering hfCas12Max and sgRNA targeting the SA or SD sites of human exon 51. The results showed that in-vivo hfCas12Max gene editing effectively restored dystrophin protein expression in the humanized DMD mouse model and improved the pathological features of DMD, including the histopathology and grip strength of the mouse model. Additionally, the all-in-one AAV particles induced therapeutically acceptable editing efficiency in exon 51 SD in healthy wild-type monkeys at a low dose with no hfCas12Max-related toxicity, indicating a good safety profile. Our research is a significant step toward the therapeutic application of gene editing correction for patients with DMD.