OpenAI o3

A model that specializes in areas such as mathematics, coding, and science. The o3-mini model supports structured outputs, function calling, developer messages, and streaming. It offers three adjustable reasoning levels, letting you balance speed against depth. The model achieves results comparable to o1 at lower latency and outperforms o1-mini on advanced STEM tasks. The maximum number of input tokens is 200,000 and 100,000 for output. Training data goes up to October 2023.