According to the company, o3 is considered the most advanced thinking model today, demonstrating outstanding performance in the fields of programming, toanistics and science.
Meanwhile, o4-mini is a more cost-effective version but still brings impressive results in similar fields.
The outstanding breakthrough of these two models is the ability to use and combine all available tools in ChatGPT, including web browser, image creation, file processing and source code analysis.
This is the first time an OpenAI language model has been able to solve multi-step complex problems by acting more independently and efficiently.
In addition, o3 and o4-mini also have improved vision: not only recognizing images, but also being able to interpret and "think" about visual content such as whiteboard, diagram, hand drawings - even when the image quality is poor.
These models can even edit images during the thinking process to come up with a suitable solution.
OpenAI emphasizes: The combination of top-notch thinking capabilities and full access to tools creates a leap in performance across both academic standards and practical work, setting a new standard for intelligence and usefulness.
Currently, users of ChatGPT Plus, Pro and Team packages can experience o3 and o4-mini. In the coming weeks, OpenAI will continue to launch the o3-pro, an even more powerful version for Pro users.
