Openi has launched A more powerful version of his O1 “reasoning” AI model, O1-Pro, in his developer API.
According to OpenAi, O1-Pro uses more computing than O1 to offer ‘consistently better reactions’. Currently it is only available to select developers – those who have spent at least $ 5 on OpenAI API services – and it is pricey. Very pricey.
OpenAi charges $ 150 per million tokens (~ 750,000 words) in the model and $ 600 per million tokens generated by the model. That is twice the price of the GPT-4.5 from OpenAI for input and 10x the price of regular O1.
OpenAI bets that the improved performance of O1-Pro developers will convince to pay those princely amounts.
“O1-Pro in the API is a version of O1 that uses more computing to think harder and to offer even better answers to the most difficult problems,” a OpenAi spokesperson told Techcrunch. “After we have received many requests from our developer community, we are delighted to bring it to the API to offer even more reliable answers.”
But early impressions of O1-Pro, which has been available since December in OpenAI’s AI-driven chatbot platform Chatgpt for Chatgpt Pro subscribers, were not incredibly positive. The model struggled with Sudoku puzzles, found users and were activated by simple optical illusion jokes.
Moreover, certain OpenAI showed internal benchmarks at the end of last year that O1-Pro performed slightly better than the standard O1 on coding and mathematical problems. It did answer those problems more reliable, but the benchmarks found.