Generative artificial intelligence (GenAI) is rapidly reshaping higher education and students’ academic writing practices. However, existing research has largely focused on technology acceptance and usage intention, with less attention to whether and how GenAI use can be sustained. This study investigates the determinants and dimensions of sustainable GenAI use among university students through a mixed-methods approach, combining an online survey of 1,084 Chinese university students with semi-structured interviews. An extended Unified Theory of Acceptance and Use of Technology (UTAUT) model was employed to examine the quantitative relationships. The results indicated that performance expectancy (β = 0.195, p < 0.001), effort expectancy (β = 0.094, p = 0.030), social influence (β = 0.147, p < 0.001), perceived enjoyment (β = 0.224, p < 0.001), and perceived creativity support (β = 0.211, p < 0.001) positively affected behavioral intention, whereas perceived risk had a negative effect (β = −0.044, p = 0.019). Facilitating conditions (β = 0.355, p < 0.001) and behavioral intention (β = 0.531, p < 0.001) were positively associated with actual usage behavior. Additionally, gender, educational level, usage experience, and disciplinary background exerted significant moderating effects. Qualitative findings identified three interconnected dimensions of sustainable GenAI use: long-term continuance, balanced and moderate use, and responsible and ethical use. By extending UTAUT beyond technology acceptance toward a multidimensional conceptualization of sustainable use, this study showed that sustainable GenAI use depended not only on technology acceptance but also on students’ ability to engage with GenAI critically, responsibly, and autonomously. The findings provide practical insights for fostering AI literacy and the appropriate integration of GenAI in higher education.