These models differ from previous ones by being able to think through a question and analyze it logically before responding. This feature allows them to provide more accurate, reasoned, and reliable answers. According to OpenAI, o3 is the most powerful AI model they have released so far. It surpasses previous models in areas such as mathematics, programming, reasoning, science, and visual understanding. Meanwhile, o4-mini is a more balanced model — considered more affordable in terms of price, speed, and performance. It is mainly designed for developers looking to integrate AI into their applications.
One of the most notable features of these models is their ability to “think with images.” Users can upload images to ChatGPT — whether a graph, diagram, or a section from a PDF — and the system will analyze these visuals and use them to form responses. It can even interpret blurry or low-quality images and work with visual objects.
In addition, the new models can write and execute Python code directly in the browser via ChatGPT, and they can perform web searches to access up-to-date information. These capabilities offer significant advantages for both everyday users and developers.
o3, o4-mini, and its more advanced version, o4-mini-high, are now available to ChatGPT Pro, Plus, and Team subscribers. These models can also be accessed via OpenAI's API platform.
As for pricing, despite its improved performance, the o3 model is offered at a reasonable rate:
$10 per million input tokens
$40 per million output tokens
The o4-mini model is priced the same as the earlier o3-mini model:
$1.10 per million input tokens
$4.40 per million output tokens
OpenAI CEO Sam Altman stated that o3 and o4-mini may be the last standalone reasoning models for ChatGPT before the release of GPT-5. The company plans to introduce a next-generation AI model that will combine both traditional GPT models and reasoning-based models into one unified system.
