OpenAI has launched o1-pro, its most advanced artificial intelligence model, which developers can already find in the company’s API.
OpenAI has launched o1-pro, its most advanced artificial intelligence model, which developers can already find in the company’s API.
«O1-pro in the API is a version of o1 that uses more computing resources to think better and provide even better answers to the most complex problems,» writes TechCrunch, citing an OpenAI representative.
He added that, following requests from the developer community, the company has moved o1-pro to an API «to offer even more reliable answers.»
The OpenAI o1-pro price for 1 million input tokens is $150 and $600 for output. This is twice as expensive as OpenAI GPT-4.5 for input and 10 times more expensive than regular o1.
Announcing the price in X, OpenAI tweeted: «It uses more compute than o1 to provide consistently better answers.»
The price of an API for AI models is measured by the amount of data (words, images, video, audio) you feed into the model and how much data you get back. Output costs more than input.
A common measure for this is 1 million «tokens.» In AI parlance, a «token» is like an atomic unit of data. When text is fed into a model, words and sentences are broken down into these tokens for processing, which can be multiple letters long. For OpenAI models, one token is roughly four characters in English. So a paragraph is roughly 100 tokens, give or take.
OpenAI announced a new subscription for the o1 pro version in early December 2024. The model was available on the ChatGPT chatbot platform based on OpenAI artificial intelligence for ChatGPT Pro subscribers. At the time, a company representative wrote that compared to o1 and o1-preview, the o1 pro mode performs better on complex machine learning tests in mathematics, natural sciences, and coding. After testing the new product, AI researcher and co-founder of OpenAI Andrey Karpaty reported that o1-pro is ahead of DeepSeek-R1. However, users were not thrilled with the model, noting that o1 pro could not cope with Sudoku puzzles and made mistakes due to simple jokes about optical illusions.
