AI Mathematical Olympiad Solver

This research presents AIMOS (AI- Powered Intelligent Math Operations System) — a multi- LLM orchestration framework to improve the correctness and reliability of mathematical problem- solving tasks. The framework annotates input questions to a category of mathematical discourse — Algebra, Calculus, Geometry, Trigonometry, or Probability & Statistics, then routes the questions to multiple large language model (LLM) APIs (including GPT-4, Claude, Gemini and Mistral) and levers their responses. The consensus algorithm and the trust-based algorithm follow to select the final response output with further symbolic verification of mathematical correctness. This experimental framework demonstrates that AIMOS outperforms use of single LLMs, with respect to correctness, reliability, and interpretability, and can be used for educational, research, and analytical purposes.