OpenAI has made the “o3-pro”, an improved version of the “o3” artificial intelligence model announced in the first half of 2025, available to users. The new model is described as the most capable artificial intelligence the company has developed to date.
OpenAI has launched the o3-pro model
O3-pro became active for ChatGPT Pro and Team users as of Tuesday, June 10. It is planned to be available to Enterprise and Edu users within a week. It was also made available on the same day via the developer API.
O3-pro replaces the current “o1-pro” model. The model is in the “reasoning” class, which stands out with its problem-solving skills by thinking step by step, especially in areas such as physics, mathematics and software. Such models can produce more stable and reliable outputs compared to traditional artificial intelligence approaches.
According to OpenAI’s statement, o3-pro shows superior performance compared to previous versions, especially in areas such as science, education, programming, business world and writing support. The model’s outputs are said to have received very high scores in terms of clarity, comprehensiveness, compliance with instructions and accuracy.
This model not only responds better to written commands, but also has the capacity to produce user-specific responses. It can provide personalized responses using its memory feature. It can also run Python, perform logical analysis on visual data, examine file content and search the internet.
The pricing determined for use via the API has also been announced. $20 is requested for one million input tokens and $80 for one million output tokens. The input token represents the text presented to the model; the output token represents the responses produced by the model. One million input tokens is equivalent to approximately 750 thousand words.
In performance evaluations, o3-pro also showed remarkable success when compared to other major artificial intelligence models. According to OpenAI’s data, o3-pro left Google’s most powerful model Gemini 2.5 Pro behind in the AIME 2024 test, which measures mathematical skills. In the GPQA Diamond test, which requires a doctorate level of scientific knowledge, it passed Anthropic’s Claude 4 Opus model.
It is also stated that the O3-pro has longer response times compared to previous models. It was stated that this delay is due to the model’s more in-depth and multi-stage processing.
{{user}} {{datetime}}
{{text}}