OpenAI releases o3-Pro, an improved model of o3 that thinks longer and answers more accurately
OpenAI released ' o3-pro ', an enhanced version of the inference model o3, on June 11, 2025. While based on o3, it is designed to think for longer and emphasize the accuracy of answers.
Model Release Notes | OpenAI Help Center
OpenAI releases o3-pro, a souped-up version of its o3 AI reasoning model | TechCrunch
https://techcrunch.com/2025/06/10/openai-releases-o3-pro-a-souped-up-version-of-its-o3-ai-reasoning-model/
o3 is an AI model announced by OpenAI in April 2025, and is said to be 'the most advanced inference model in OpenAI's history' because it outperformed previous models in benchmarks that measure mathematical, coding, reasoning, scientific, and visual understanding abilities.
OpenAI announces 'o3' and 'o4-mini,' which it evaluates as 'the most advanced inference model in the company's history,' enabling 'Thinking with images' that can be used to think using images as well as text - GIGAZINE
The o3-pro is a model based on the o3, and can perform web searches, analyze files, infer visual input, use Python, and personalize responses using memory. It prioritizes reliability over speed, and is designed to 'perform inference for longer and output reliable answers.' As a result, it takes longer to complete a response than the previous generation o1-pro, but OpenAI says, 'We recommend using it for difficult problems where it is worth waiting a few minutes.'
In expert evaluations, o3-pro consistently rated higher than o3 in all categories tested, especially in key areas such as science, education, programming, business, and writing assistance. o3-pro also consistently received higher marks for clarity, comprehensiveness, following instructions, and accuracy.
OpenAI also reports that o3-pro has been shown to consistently outperform both o1-pro and o3 in math, science, and coding benchmarks.
In order to evaluate the reliability more strictly, the results are as follows: o3-pro achieved the highest scores in mathematics, science, and coding.
o3-pro will be available to ChatGPT Pro and Team users, replacing the existing o1-pro. Enterprise and Edu users will be able to use it the following week. The developer API costs $20 (approximately ¥2,800) per million input tokens and $80 (approximately ¥11,200) per million output tokens.
OpenAI o3-pro is available in the model picker for Pro and Team users starting today, replacing OpenAI o1-pro.
— OpenAI (@OpenAI) June 10, 2025
Enterprise and Edu users will get access the week after.
As o3-pro uses the same underlying model as o3, full safety details can be found in the o3 system card.…
However, at the time of writing, OpenAI has temporarily disabled chat with o3-pro in ChatGPT, stating that it is 'solving technical issues.' In addition, o3-pro cannot generate images, and OpenAI's AI-powered workspace feature, Canvas, is not supported by o3-pro.
Related Posts:
in Software, Posted by log1i_yk