{"id":174277,"date":"2025-06-11T00:21:09","date_gmt":"2025-06-11T00:21:09","guid":{"rendered":"https:\/\/www.europesays.com\/uk\/174277\/"},"modified":"2025-06-11T00:21:09","modified_gmt":"2025-06-11T00:21:09","slug":"openai-releases-o3-pro-a-souped-up-version-of-its-o3-ai-reasoning-model","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/uk\/174277\/","title":{"rendered":"OpenAI releases o3-pro, a souped-up version of its o3 AI reasoning model"},"content":{"rendered":"

OpenAI has launched<\/a> o3-pro, an AI model that the company claims is its most capable yet.<\/p>\n

O3-pro is a version of OpenAI\u2019s o3<\/a>, a reasoning model that the startup launched earlier this year. As opposed to conventional AI models, reasoning models work through problems step by step, enabling them to perform more reliably in domains like physics, math, and coding. <\/p>\n

O3-pro is available for ChatGPT Pro and Team users starting Tuesday, replacing the o1-pro<\/a> model. Enterprise and Edu users will get access the week after, OpenAI says. O3-pro is also live in OpenAI\u2019s developer API as of this afternoon. <\/p>\n

O3-pro is priced at $20 per million input tokens and $80 per million output tokens in the API. Input tokens are tokens fed into the model, while\u00a0output tokens\u00a0are\u00a0tokens\u00a0that the model generates based on the input\u00a0tokens.<\/p>\n

A million input tokens is equivalent to about 750,000 words, a bit longer than \u201cWar and Peace.\u201d<\/p>\n

\n

OpenAI o3-pro is available in the model picker for Pro and Team users starting today, replacing OpenAI o1-pro.<\/p>\n

Enterprise and Edu users will get access the week after.<\/p>\n

As o3-pro uses the same underlying model as o3, full safety details can be found in the o3 system card.\u2026<\/p>\n

\u2014 OpenAI (@OpenAI) June 10, 2025<\/a><\/p><\/blockquote>\n

\u201cIn expert evaluations, reviewers consistently prefer o3-pro over o3 in every tested category and especially in key domains like science, education, programming, business, and writing help,\u201d OpenAI writes in a changelog. \u201cReviewers also rated o3-pro consistently higher for clarity, comprehensiveness, instruction-following, and accuracy.\u201d<\/p>\n

O3-pro has access to tools, according to OpenAI, allowing it to search the web, analyze files, reason about visual inputs, use Python, personalize its responses leveraging memory, and more. As a drawback, the model\u2019s responses typically take longer than o1-pro to complete, according to OpenAI. <\/p>\n

O3-pro has other limitations. Temporary chats with the model in ChatGPT are disabled for now while OpenAI resolves a \u201ctechnical issue.\u201d O3-pro can\u2019t generate images. And Canvas, OpenAI\u2019s AI-powered workspace feature, isn\u2019t supported by o3-pro.<\/p>\n

On the plus side, o3-pro achieves impressive scores in popular AI benchmarks, according to OpenAI\u2019s internal testing. On AIME 2024, which evaluates a model\u2019s math skills, o3-pro scores better than Google\u2019s top-performing AI model, Gemini 2.5 Pro<\/a>. O3-pro also beats Anthropic\u2019s recently released Claude 4 Opus<\/a> on GPQA Diamond, a test of PhD-level science knowledge.<\/p>\n