In recent years, there has been a significant increase in the use of Artificial Intelligence (AI) in digital language learning. However, research on investigating AI’s role in language learning among young learners is limited. This study is based on the design and implementation of a self-developed AI-enhanced Chinese vocabulary learning system called ARCHe for young learners. The research involved 140 primary 2 students from the three primary schools in Singapore, using the system for 8–10 weeks. Findings show that using ARCHe improved students’ learning performances overall, regardless of school context or prior language proficiency. Students accepted ARCHe positively, with higher language proficiency associated with greater ease of use perceptions, while perceived usefulness and enjoyment were consistent across different levels of language proficiency. The study suggests providing additional support for learners with lower language proficiency, potentially through multimodal feedback. Future research could further explore students’ epistemic emotions beyond acceptance.