OpenAI has officially launched its latest AI reasoning model, o3-mini, inside ChatGPT and its API services. For the first time, a version of this model with rate limits is available to free ChatGPT users. This release comes exactly two weeks after CEO Sam Altman hinted that o3-mini would arrive “in a couple of weeks.”
Introduced initially during OpenAI’s “12 Days of Ship-mas” in December, o3-mini is designed to match the performance of o1 in areas like mathematics, coding, and science while offering significantly faster response times. OpenAI claims that o3-mini is 24% quicker than o1-mini and provides more accurate answers. Like its predecessor, the model explains how it derives answers rather than simply presenting results.
Expanded access and enhanced features
In benchmark tests conducted by OpenAI, the o3 series outperformed the o1 model across various reasoning and coding tasks. The o3-mini version now brings these benefits at a lower cost and reduced latency. Developers can access o3-mini through OpenAI’s API services, including the Chat Completions API, Assistants API, and Batch API.
Paid users will have access to o3-mini-high, which OpenAI describes as the “best coding option in ChatGPT.” This version provides more intelligent responses, though they take slightly longer to generate. Additionally, o3-mini supports web search functionality, allowing users to find answers with linked sources.
Free access and integration with Microsoft services
This launch marks the first time OpenAI’s reasoning models are available to free ChatGPT users. It comes just days after Microsoft made o1 free for all Copilot users and DeepSeek introduced its latest AI advancements. Free users can try o3-mini by selecting ChatGPT’s “Reason” feature, with rate limits similar to GPT-4o.
Meanwhile, ChatGPT Plus, Team, and Pro users will also get access to o3-mini starting today. OpenAI is tripling the daily message limit for Plus and Team users to 150 messages, while Pro users, who pay US$200 per month, will have unlimited access.
OpenAI’s o3-mini is also being integrated into Microsoft Azure OpenAI Service, GitHub Copilot, and GitHub models, expanding its availability across Microsoft’s ecosystem.
