AI BriefingOpenAI o3-mini
Key point
OpenAI has released o3-mini, a highly efficient reasoning model optimized for STEM fields.
Details
OpenAI unveiled o3-mini, a new model in its reasoning series. This model is notable for delivering excellent performance in STEM fields such as science, math, and coding, while maintaining the low cost and low latency of o1-mini.
It is designed to be immediately applicable in production environments by supporting the following key features for developers:
- Support for Function calling, Structured Outputs, and developer messages
- Streaming support
- Optimization of reasoning intensity via Reasoning effort (low, medium, high) options
It is immediately available to both ChatGPT users and API users. For ChatGPT Plus and Team users, the daily message limit has been tripled from the existing 50 to 150. In addition, Free plan users can now use a reasoning model for the first time via the 'Reason' option.
In terms of performance, o3-mini (medium effort) shows performance on par with o1 in math, coding, and science. In particular, it records excellent results on high-difficulty reasoning evaluations such as AIME and GPQA, reducing errors by 39% compared to o1-mini and providing clearer answers.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
OpenAI Prepares New $500/Month ChatGPT Pro Max Plan with 'Fastest' Inference
Class Action Suit Alleges OpenAI Routed ChatGPT Conversations to Human Reviewers Without Clear DisclosureOpenAI Significantly Expands Influencer Marketing to Enhance ChatGPT's Image