After the unprecedented agreement between Openai and Google, now the first one just had just launch O3-Proa model of AI that the company cries out that it is the most capable so far.
O3-PRO is one OPENAI O3 Reasoning Model version that the Startup He launched earlier this year. A type of model that, vice versa that conventional AI, develops the problems step by step.
This allows you to perform much better in subjects such as physics, mathematics and coding. A leap forward that hits chatgpt to duel with Gemini 2.5 Pro updated days ago and Claude 4 Opus of Anthropic.
O3-PRO is now available for users who pay for the Chatgpt Pro and Team Subscription Since yesterday, and replaces the O1-Pro model. Enterprise and Edu users will have access next week.
The API for developers It has already has O3-Pro for hours. O3-PRO has a cost of $ 20 for one million entrance tokens and $ 80 per million departure tokens in the API.
OPENAI O3-PRO IS AVAILABLE IN THE MODEL PICKER FOR PRO AND TEAM USERS STARTING ALL, REPLACING OPENAI O1-PRO.
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 …
– OpenAi (@openai) June 10, 2025
To better understand these figures, one million tokens equivalent to about 750,000 wordsa little more than a novel like War and peace From Lev Tolstob. IA companies charge third parties for the generation of the text through these tokens.
And now the interesting The O3-Pro performance. OpenAI maintains that, according to experts’ evaluations, they prefer O3-Pro on O3 in any proven category, and especially in key matters such as science, education, programming, business and writing.
This is O3-Pro in the face of other OpenAI models
There is another key point in O3-Proand it is your access to tools such as web search, file analysis, reasoning about visual entries, is capable of using Python, customizes your answers and more.
There are several limitations In the new experience according to the Startup from Your website. One is that the model takes a little longer to respond than O1-Pro, as Openai maintains.
Another is that temporary chats are deactivated while a technical problem is solved. And both the generation of images with AI and the use of Canvas, their work space function, are not supported by O3-Pro.
The next question we ask ourselves is How to pay other models In order to really fit your capacity: O3-Pro achieves impressive scores in the Benchmarks of the most popular.
In Aime 2024, according to Techcruncha test capable of evaluating the mathematical skills of the model, O3-Pro score better than Gemini 2.5 ProGoogle’s top model. O3-PRO also beat Claude 4 opus in GPQA Diamond.
We still do not have a comparative table in which we can appreciate the differences with the mentioned AI models as well as the latest Deepseek R1 update that arrived a few weeks ago.
