HIGHLIGHTS

OpenAI has introduced two innovative AI reasoning models: o3 and o4-mini.

These cutting-edge models are engineered to pause and meticulously analyze questions before delivering their responses.

The o3 model stands as OpenAI’s most advanced reasoning model to date.

OpenAI has recently announced the release of two groundbreaking AI reasoning models: the o3 model and the o4-mini model. These advanced models are designed with the capability to pause and thoroughly evaluate questions before rendering their answers. According to OpenAI, these releases mark the pinnacle of their AI development, offering smarter solutions than any prior models they have put forth.

The o3 model is touted as OpenAI’s most robust reasoning model yet, showcasing superior performance across various domains including mathematics, coding, logical reasoning, scientific inquiry, and visual comprehension. Notably, it excels in coding assessments, achieving a remarkable score of 69.1% on the SWE-bench verified test, showcasing a significant leap in capability compared to its predecessors.

On the other hand, the o4-mini model strikes an optimal balance between cost, speed, and performance. OpenAI highlighted that “o4-mini is a more compact model specifically refined for quick, economical reasoning—it provides impressive performance given its size and cost, especially in mathematical tasks, coding challenges, and visual assignments,” as articulated in their recent blog post.

Curious readers can find additional insights by checking out this related article: OpenAI is developing its own social media platform to compete with Elon Musk’s X, according to reports.

A standout feature of both the o3 and o4-mini models is their capability to generate responses by utilizing every available tool within ChatGPT. This includes functions such as web browsing, executing Python code, and image processing. Notably, these models mark the first time OpenAI has enabled AI reasoning involving images.

As highlighted by the company, “Users can upload photographs of whiteboards, diagrammatic representations from textbooks, or even rough sketches. The model is equipped to interpret these images, regardless of if the images are unclear, reversed, or of low quality. Moreover, with the utilization of tools, the models can dynamically manipulate images—rotating, zooming, or altering them as part of their reasoning process,” the company explained.

For those eager to learn more, another informative article can be found here: OpenAI has rolled out new GPT-4.1 AI models: all the details you need to know.

Availability

As of today, users of ChatGPT Plus, Pro, and Team can access o3, o4-mini, and o4-mini-high within the model selection area, effectively replacing the previous versions: o1, o3-mini, and o3-mini-high. Additionally, ChatGPT Enterprise and Edu users are expected to gain access within a week. Free-tier users will have the opportunity to experiment with o4-mini by choosing ‘Think’ in the composition tool prior to submitting their queries.

Looking ahead, OpenAI plans to introduce the o3-pro model. With these new offerings, OpenAI aims to continue pushing the envelope of what is achievable in AI technology.