OpenAI has released a more powerful version of its o1 AI model, called o1-pro, and added it to its developer API. The company says o1-pro uses more computing power than o1 to consistently produce better answers. o1-pro is currently only available to select developers — those who have already spent at least $5 on OpenAI’s API services.
Image source: OpenAI
OpenAI charges $150 per million tokens fed to the AI model for processing (approximately 750,000 words) and $600 per million tokens generated by the AI model. This is twice as expensive as GPT-4.5 for processing input data, and 10 times more expensive than the “reasoning” AI model o1 for generating text.
OpenAI hopes that the o1-pro’s increased performance will convince developers to pay such hefty sums. The company claims that the new product uses additional computing resources to “think harder” and provide more accurate answers to complex questions.
«”O1-pro in the API is a version of o1 that uses more computing power to think even deeper and give even more accurate answers to the most complex problems,” an OpenAI spokesperson said. He added that the company decided to include the model in the API after numerous requests from developers.
However, the first reviews of the AI model o1-pro, which has been available on the OpenAI AI chatbot platform for ChatGPT Pro subscribers since December 2024, were not very positive. Users found that the newcomer had difficulty coping with Sudoku puzzles, and simple jokes about optical illusions baffled it.
Moreover, some internal tests conducted by OpenAI in late 2024 showed that the o1-pro only slightly outperformed the o1 in programming and math tasks, but showed more consistent results.