Diabetes is a chronic metabolic disorder that requires continuous monitoring and personalized management to prevent severe complications. In this study, we introduce DiabetesLLM, a French-language model designed to answer diabetes-related medical questions and provide patient support. Leveraging advanced natural language processing techniques, DiabetesLLM was fine-tuned on a comprehensive, domain-specific dataset containing medical questions paired with context-aware answers relevant to diabetes care. Experimental evaluations show that DiabetesLLM delivers highly accurate and clinically relevant responses, outperforming baseline multilingual and generalist models. Beyond its technical performance, the model enhances patient engagement by providing timely, personalized guidance in French, addressing a critical gap in digital health tools for Francophone populations. Fine-tuned large language models like DiabetesLLM have the potential to improve diabetes management through AI-driven conversational agents, contributing not only to better health outcomes but also to greater patient empowerment in chronic disease care.