As artificial intelligence (AI) has been integrated into foreign language (FL) education, learners' well-being is influenced by various factors, including technological, personal and contextual elements. However, few studies explored how external and internal factors jointly shape FL learners' well-being in the era of generative AI. To fill this gap, this study explores the effects of generative AI acceptance, perceived teachers' enthusiasm and self-efficacy on FL learners' well-being by investigating 613 university learners of English as a foreign language (EFL). The structural equation modelling results reveal that (1) generative AI acceptance positively predicts EFL learners' well-being and self-efficacy; (2) perceived teachers' enthusiasm does not predict learners' well-being and positively predicts EFL learners' self-efficacy; and (3) the self-efficacy for receptive skills mediates the relationship between generative AI acceptance/perceived teachers' enthusiasm and EFL learners' well-being, whereas self-efficacy for productive skills does not play the mediation role. This research broadens the understanding of the antecedents of EFL learners' well-being and extends the application of self-efficacy theory in the AI-driven educational environment, providing significant pedagogical implications.