Technology

o1-mini

OpenAI's o1-mini is the specialized, cost-efficient Large Language Model (LLM) engineered for advanced, step-by-step reasoning in STEM applications.

This is OpenAI’s high-performance, budget-friendly reasoning model, launched in September 2024. The o1-mini excels at complex problem-solving, particularly in technical domains like mathematics and coding, achieving a 70.0% score on the AIME high school math competition and a 1650 Elo rating on Codeforces. It utilizes the same high-compute reinforcement learning (RL) pipeline as the larger o1 model but is optimized for speed and cost-efficiency, making it up to 80% cheaper than the o1-preview model for API users. Deploy o1-mini for applications demanding deep, systematic reasoning without the overhead of broad world knowledge.