OpenAI has released o3-mini, a new model designed for tasks requiring logical reasoning. Now available in ChatGPT and the API, it offers enhanced accuracy and speed for solving problems in science, mathematics, and programming. This version replaces o1-mini, delivering better performance with lower latency.
The model supports function calling, structured outputs, and different levels of reasoning effort, allowing users to tailor its performance to specific needs. In ChatGPT, o3-mini operates with medium reasoning effort, while Plus, Team, and Pro subscribers can switch to o3-mini-high, which offers advanced intelligence capabilities.
Free users can now access the model by selecting the "Reason" mode in the chat. OpenAI has also increased the daily message limit for Plus and Team users from 50 to 150, while Pro users enjoy unlimited access. Additionally, o3-mini is integrated with search, providing real-time answers with source links.
Expert evaluations show that o3-mini is 39% more accurate on complex questions than its predecessor, with users preferring its responses 56% of the time. This model makes advanced AI reasoning tools more accessible and efficient for both developers and a broader audience.