Openai has announced the launch of the O3-PRO, an improved variant of its star reasoning model that uses more computational resources to think more intensity and provide consistently better answers.
Unlike conventional AI models, reasoning models solve step by step, which allows them a more reliable performance in areas such as physics, mathematics and programming. In return, they use more resources and take longer to issue the answers, but they are generally better in advanced projects.
In expert evaluations, reviewers systematically prefer O3-PRO on O3 in all categories evaluated, especially in the key areas cited with science, education, programming, business or help with writing. The reviewers also systematically described O3-PRO with a higher score in terms of clarity, exhaustivity, monitoring of instructions and precision.
In order for all customers to make the most of the model, O3-PRO admit the call to tools. Thus, according to the user’s application, the model can perform web searches, analyze files, reason about visual entries, use languages such as Python, customize answers using memory and more. This is a great improvement with respect to O1-PRO, which lacked access to tools. However, the disadvantage is the aforementioned: O3-Pro takes longer to respond. OpenAI recommends using O3-Pro when reliability is more important than speed.
AVAILABILITY AND Price OF THE O3-PRO
The new version It is now available In the chatgpt model selector for Pro and Team versions. Chatgpt Enterprise and Education users will have access to this new model next week.
For developers, the O3-PRO model has a price of $ 20 per million input tokens and $ 80 per million departure tokens. OpenAI recommends developers to use the background mode with O3-PRO to avoid waiting times. The model has the deadline of knowledge on May 31, 2024.
The OpenAI team has indicated the current limitations that will be resolved in the upcoming dates, such as the disabling of temporary chats, the generation of images or the use of Canvas.
