OpenAI releases o3-pro, a souped-up version of its o3 AI reasoning model

Openai has has initiated O3-PRO, the company’s claiming AI model is still the most skilled.

O3-PRO is a version of Openai O3A thought model of starting this year before the beginning. Unlike ordinary AI models, justification models work through step-by-step problems, allow you to perform more reliably in domains such as physics, math and coding.

O3-Pro is available for Chatgpt Pro and Team users starting on Tuesday O1-PRO The model. Enterprise and Edu users will get access to a week later. The O3-Pro also lives in the API of the Developer of this afternoon.

O3-Pro is worth $ 80 in $ 20,000 in one million input versions and $ 80 in API. Input signs are fed to the model, the Tokens in the output verses are Token, which is based on the model entry verses.

One million input verse is about 750,000 words, “War and Peace” a little longer than.

“In expert assessments, reviewers consistently prefer each test, such as every test, education, programming, work and writing, especially in the O3 Pro in the Key Domains,” he said. “Reviewers also rated in a consistent O3-Pro in a consistent for clarity, comprehensively, instructions and accuracy.”

According to O3-Pro, WePenaI, use the answers to analyze the webs, use the information, using PYTHON, using the answers, use their answers and personalize more. As a drawback, the model’s answers, according to Openai, usually take longer than O1 Pro to complete.

O3-PRO has other restrictions. Temporary conversations with the model in Chatgpt, the Openai “technical issue” is still disabled so far. O3-PRO can’t create images. And the canvas, open AI-powered Work Space feature is not supported by O3-Pro.

Moreover, O3-Pro, according to the internal test of Openai, obtains effective scores in popular AI pubs. In 2024, which evaluates the math skills of a model, O3-Pro scores better than the best performed AI model, Gemini 2.5 Pro. O3-Pro also beats anthropik’s recently released Close the 4th job In GPGA Diamond, a test of doctor of philosophy scientific knowledge.