The API is available to select developers on tiers 1–5
OpenAI says the o1-pro model generates “consistently better answers”
It works with vision, function calling, and structured outputs
The o1-pro AI model offers a context window of 2,00,000 tokens
Photo Credit: Reuters
OpenAI released a new application programming interface (API) for the recently introduced o1-pro artificial intelligence (AI) model on Wednesday. The new API will offer developers better responses than any of the existing models by the AI firm by using more compute. However, the company is passing through the higher computational costs back to the developers by making the API its most expensive offering till date. Notably, ChatGPT Pro subscribers can access the o1-pro mode with rate limits without paying any extra charges.
OpenAI's o1-Pro Gets an API
In a post on X (formerly known as Twitter), the official handle of OpenAI Developers announced the launch of the o1-pro API. The AI firm claimed that it was released after multiple developers requested the company to make the model's capabilities available via an API. The model uses more compute compared to the o1 model to generate “consistently better responses.”
However, since it was so far available as a feature within ChatGPT, its benchmark scores are not known. OpenAI said that the API comes with support for vision, function calling, and structured outputs. It also works with Responses and Batch APIs. The model supports text and images as input, but audio input is not supported. The model can only generate text output.
As per the company's model document, the o1-pro has a context window of 2,00,000 tokens, a knowledge cutoff of October 2023, and reasoning token support. However, despite the capabilities, the API is also the AI firm's costliest offering. It is priced at $150 (Rs. 12,900) per million input tokens and $600 (roughly Rs. 51,800) per million output tokens. Currently, OpenAI has made it available to select developers on the paid 1-5 tiers.
To draw a parallel, the o1-mini and o3-mini models cost $1.10 (roughly Rs. 94) per million input tokens and $4.40 (roughly Rs. 380) per million output tokens. Moreover, developers should also expect additional costs with the o1-pro as it is a reasoning model and API users are charged for the reasoning tokens. In the past, several developers have raised concerns over the fact that they cannot see the chain of thought (CoT).
ChatGPT Pro subscription, which costs $200 (roughly Rs. 17,270) a month, offers access to the o1-pro model via a mode created within the platform. It comes with rate limits but does not cost users any extra amount.
OpenAI Introduces New Audio Models in API, Can Be Used for Agentic Workflows21 March 2025
Search
