OpenAI Launches New AI Models o3 and o4-mini for Enhanced Reasoning

OpenAI has introduced two innovative artificial intelligence models, o3 and o4-mini, aimed at improving AI’s reasoning capacities and its ability to interpret and manage images during problem-solving tasks.

Overview of the New Models

Recognised as the company’s “most powerful reasoning model,” o3 holds the top position within OpenAI’s current model hierarchy. Alongside it is o4-mini, a more compact model that the company asserts “delivers impressive performance relative to its size and cost,” as stated in an official blog announcement.

Image Integration in Reasoning

Both models are engineered to “think” visually, incorporating imagery directly into their reasoning processes. This enables them to examine, magnify, or manipulate images to enhance their analysis. Such capabilities could significantly benefit tasks related to diagrams, sketches, or whiteboard content, marking a considerable advancement in the integration of text and visual information.

Access to Advanced Tools for Users

OpenAI has also revealed that these models will be equipped with complete access to the range of ChatGPT tools. These features include web browsing, file analysis, code generation, and image creation functionalities that are generally available only to premium users of the chatbot service. Effective immediately, subscribers of ChatGPT Plus, Pro, and Team tiers can utilise these tools through o3, o4-mini, and o4-mini-high. Access to the more advanced o3-pro version is anticipated to be available “in the coming weeks.”

Retirement of Older Models

In conjunction with this announcement, OpenAI will start phasing out older models such as o1, o3-mini, and o3-mini-high from its premium offerings.

Recent Updates to OpenAI’s Model Portfolio

This launch coincides with OpenAI’s recent unveiling of GPT-4.1, the newest version of its prominent generative model, highlighting a week filled with significant enhancements across its model lineup.