OpenAI has officially launched its o3-mini reasoning model, fulfilling CEO Sam Altman’s promise from two weeks ago. The latest model is now available in ChatGPT and OpenAI’s API services. For the first time, free users of ChatGPT can access a version of OpenAI’s reasoning models with rate limits.
Faster, More Accurate AI Responses
Originally introduced during OpenAI’s December “ship-mas” event, o3-mini is designed to match o1’s performance in math, coding, and science. However, it offers faster response times. OpenAI states that o3-mini responds 24% faster than o1-mini while improving accuracy. Similar to o1-mini, this model also explains how it arrives at an answer instead of just providing a result.
Enhanced Performance and API Access
Early benchmarks from OpenAI show that o3-mini outperforms o1 in reasoning and coding tasks. Developers can integrate the model via OpenAI’s API services, including:
- Chat Completions API
- Assistants API
- Batch API
Paid users will have access to o3-mini-high, which is optimized for coding and provides more advanced responses. Although it takes longer to generate answers, it offers enhanced intelligence and deeper reasoning.
Availability for Free and Paid Users
This marks the first time OpenAI has made a reasoning model available to free ChatGPT users. Just days after Microsoft made o1 free for Copilot users, OpenAI is expanding access by allowing free users to test o3-mini under rate limits. ChatGPT Plus, Team, and Pro users can also access the model starting today.
OpenAI is increasing message limits for Plus and Teams users to 150 messages per day. However, Pro users, who pay $200 per month, will enjoy unlimited access.
Integration with Microsoft and GitHub
Microsoft has also integrated the o3-mini reasoning model into its services. It is now available in:
- Microsoft Azure OpenAI Service
- GitHub Copilot
- GitHub models
With this latest release, OpenAI continues to push the boundaries of AI reasoning while expanding accessibility to more users.