ChatGPT, as an AI (artificial intelligence) language model, shows significant potential in the field of language teaching. However, research on teachers’ acceptance of ChatGPT-assisted language teaching remains limited. This study aims to explore teachers’ acceptance intention of ChatGPT-assisted language teaching by incorporating the Attributes of Innovation (AOI) theory into the Technology Acceptance Model (TAM). The factor “trialability” in the AOI theory was removed during the framework establishment phase because ChatGPT allows users to use it for free, enabling them to directly access and evaluate ChatGPT’s performance, thus reducing the necessity for trialability. We ultimately validated four factors in the AOI theory and three factors in the TAM, and proposed 11 hypotheses. We distributed 229 questionnaires to English-as-a-foreign-language (EFL) student teachers, obtained 187 valid responses, and analyzed the data using the structural equation modeling techniques via the software Partial Least Squares (PLS). Finally, nine out of 11 hypotheses were confirmed, and the unconfirmed two hypotheses were “compatibility (CPA) positively impacts perceived usefulness (PU)” (H6) and “compatibility (CPA) positively impacts perceived ease of use (PEoU)” (H7). The results suggested that while the Attributes of Innovation (AOI) theory is broadly applied, the impact of specific dimensions (“compatibility” in the current study) may vary across different contexts.