OpenAI has made a major announcement — the company has introduced two new generative AI models that impress not only with their text skills but also with visual capabilities. The o3 and o4-mini models are already available to ChatGPT users with Plus, Pro, and Team subscriptions, as well as to developers via the API. O3 is positioned as the company’s most powerful model for tasks requiring logic, mathematics, coding, and deep analysis, while o4-mini offers speed and low cost — the perfect balance for those seeking efficiency without extra expenses.
Introducing OpenAI o3 and o4-mini—our smartest and most capable models to date.
— OpenAI (@OpenAI) April 16, 2025
For the first time, our reasoning models can agentically use and combine every tool within ChatGPT, including web search, Python, image analysis, file interpretation, and image generation. pic.twitter.com/rDaqV0x0wE
The real breakthrough is the new models’ ability to “think with images” — now they not only see but also analyze visuals as part of their reasoning process. Users can upload even low-quality or blurry pictures, and the model will recognize, process, scale, or rotate them to find the right answer. This opens up new possibilities for solving complex problems where both text and visual information matter.
The o3 and o4-mini models can independently use all ChatGPT tools — from web search and image generation to running Python code and processing files. This multifunctionality allows them to efficiently solve complex issues by combining different technologies for the best result. This is especially evident when analyzing complex diagrams, searching for information online, or creating new images on demand.
OpenAI emphasizes that the models achieved record results in programming tests, scoring over sixty-nine percent on challenging benchmarks, significantly outperforming the company’s previous developments. For developers, these models are available at an attractive price — o3 costs ten dollars per million input tokens, while o4-mini retains the price of the previous version.
A more powerful version of o3 is expected to be released in the coming weeks — exclusively for ChatGPT Pro subscribers. According to OpenAI CEO Sam Altman, these models may be the last standalone AI models for logical reasoning in ChatGPT before the next generation — GPT-5 — arrives.
