OpenAI releases o3-mini, a cost-efficient reasoning model for STEM and coding

OpenAI has released OpenAI o3-mini, a new small reasoning model available in ChatGPT and the API, designed to deliver exceptional STEM capabilities with low cost and reduced latency. It supports developer features like function calling and structured outputs, and replaces OpenAI o1-mini for Plus and Team users.

• Matches OpenAI o1 performance on AIME and GPQA evaluations while delivering responses 24% faster than o1-mini.
• Offers three reasoning effort options (low, medium, high) to balance speed and accuracy.
• Increases rate limits for Plus and Team users from 50 to 150 messages per day.
• Available to free plan users in ChatGPT for the first time with a reasoning model.

The release aims to make high-quality AI more accessible by optimizing reasoning for technical domains while maintaining safety and efficiency.