OpenAI releases o3 and o4-mini AI models with native agentic abilities

OpenAI highlighted that these new o-series reasoning models are the first to include images directly in their chain of thought

OpenAI has released two new o-series reasoning models, namely o3 and o4-mini. The company said these are its first AI reasoning models to have built-in agentic abilities, allowing them to combine and use tools available within ChatGPT, such as web searching, file analysis, visual reasoning, and image generation. OpenAI noted that these models are trained to decide when and how to use these tools to generate detailed responses in the correct format.

 

In a blog post, OpenAI explained that the new models use their agentic abilities to tackle complex, multi-step questions more effectively. The company sees this as an initial step toward building a true agentic ChatGPT that can independently perform tasks on a user's behalf.

 

 

OpenAI’s new agentic reasoning models: Details

• o3 model: OpenAI described the new o3 model as its most powerful reasoning model to date, doing well across domains such as coding, mathematics, science, and visual understanding. The company said o3 is ideal for solving multi-layered problems where answers are not immediately obvious. It is also said to show strong performance in visual tasks like analysing images, charts, and diagrams.
• o4-mini model: The o4-mini is a smaller model optimised for speed and cost-efficiency. Despite its size, it reportedly performs excellently in maths, coding, and visual reasoning during internal tests. OpenAI said it outperforms its predecessor, the o3-mini, in areas like data science, while also supporting higher usage limits compared to the new o3.

OpenAI’s new agentic reasoning models: Visual abilities

 

OpenAI highlighted that these models are the first to include images directly in their chain of thought. According to the company, “they don’t just see an image—they think with it.”

 

Using this ability, users can upload photos of whiteboards, textbook diagrams, or hand-drawn sketches, and the models can understand them—even when the image is blurred, reversed, or low quality. With tool use enabled, the models can also change images on the fly, performing tasks like rotating, zooming, or transforming visuals as part of their reasoning process.

Availability

 

OpenAI said that o3, o4-mini, and a higher capacity variant called o4-mini-high are now available in the model selector menu on ChatGPT for Plus, Pro, and Team subscribers. However, free-tier users can try the new o4-mini model by selecting ‘Think’ in the composer before submitting their query. The company also mentioned plans to launch a new o3-pro model soon.