Recent developments in artificial intelligence (AI) have opened up new possibilities for computer-assisted language learning (CALL). At the forefront of these technological advancements is Gemini (formerly known as Bard), an AI-driven language model developed by Google. This tool has the potential to revolutionize the way grammar is taught and learned. Thus, this paper explores its pedagogical use within the context of grammar teaching and learning. The first part briefly presents Gemini’s general affordances for language learning and its limitations, such as the potential dissemination of incorrect information, an absence of emotional and cultural cognizance, and a lack of support in spoken language development. This is followed by a detailed discussion of how learners can utilize the tool’s affordances for self-directed grammar learning. This section also presents ways in which teachers can harness Gemini’s capabilities to create an engaging, responsive teaching environment, generate tailored grammar exercises, facilitate contextual grammar practice, and produce relevant supplementary study materials that promote learner autonomy and grammatical competence.