This poster presentation introduces on-device large language models (LLMs) like Jan as privacy-focused alternatives to cloud-based AI tools in education. We'll explore how on-device LLMs maintain data security while delivering AI-enhanced learning experiences, addressing privacy concerns raised by platforms like ChatGPT. Participants will learn about the capabilities of on-device LLMs, focusing on Jan, a desktop application running open-source models locally. We'll demonstrate Jan's text generation, writing assistance, and educational support features, emphasizing data privacy. Attendees will understand how on-device LLMs offer a compelling alternative to cloud-based models, enabling educators to use AI while prioritizing privacy. They'll gain strategies for integrating these tools into their teaching, contributing to a more secure and equitable AI-enhanced learning environment.