This research focuses on the challenges of using large language models (LLMs) to solve high school math problems in Vietnam. Although LLMs have advanced natural language processing, they struggle with math problem-solving. The proposed approach involves using topic classification to identify the relevant math topic, searching for similar problems in a dataset, extracting answers, and using in-context learning for further improvement. Code generation is also used to help with computational problem-solving and provide detailed explanations. By addressing the limitations of LLMs in understanding numbers, computations, and reasoning, this research aims to enhance the accuracy and effectiveness of LLM-based approaches in supporting students in their studies. These findings have the potential to advance educational technology and assist students in confidently navigating high school math.