Our project proposes a multi-LLM-based educational platform that dynamically coordinates multiple models and novel routing strategies to deliver adaptive problem solving, college choice guidance, and long-term learning support. By selecting appropriate models for different tasks, not only can the system improve response quality but also reduces unnecessary computational cost, achieving a better performance-cost trade-off and indirectly improving environmental efficiency. The project targets the growing Gaokao education market and will be developed in collaboration with a Turkish AI education platform with over 500,000 users.
