OpenAI Releases o3-Pro Reasoning-Focused AI Model, Comes With Improved Capabilities and Tool Use

OpenAI released its o3-pro artificial intelligence (AI) model on Tuesday. Dubbed as the company's most advanced reasoning model, it is only available to the most premium tier of subscribers. The o3-pro model will be available both in ChatGPT as well as via the AI firm's application programming interface (API). The model is said to be more performant compared to o3 in subject areas such as science, education, programming, data analysis, and writing. The AI model was released just hours after OpenAI CEO Sam Altman announced an 80 percent discount in the input and output prices of o3.

OpenAI Releases o3-pro For ChatGPT Pro Subscribers

In a post on X (formerly known as Twitter), the official handle of OpenAI announced the rollout of the o3-pro model. Currently it is available to only ChatGPT Pro subscribers, and to developers via an API. Enterprise and Edu users will get access to it one week later.

The o3-pro AI model will replace the o1-pro model in the model picker option for Pro subscribers. Notably, the flagship-tier reasoning model from the o3 family arrives nearly two months after the San Francisco-based AI firm introduced the baseline o3 and the o4-mini models.

OpenAI says the latest reasoning model outperforms the non-pro variant in multiple areas. Based on internal evaluations with human testers, the company said the o3-pro achieved a 66.7 percent win rate in personal writing over the o3 model. It also achieved 62.7 percent win rate against the baseline model in computer programming. Notably, the company did not share any benchmark tests comparing the model against competitors such as Gemini 2.5 Pro or Claude Opus 4.

The o3-pro AI model is also capable of tool use, and it can access tools such as web search, file analysis, computer vision with reasoning, use Python, access memory to personalise responses, and more. However, the company also highlighted that due to access to tools, the model also takes slightly longer than o1-pro to generate responses. As such, OpenAI is asking users to choose it when accuracy of an answer is more important than speed.

There are some limitations with the large language model (LLM) as well. As per the OpenAI, the o3-pro model cannot generate images and does not support temporary chats or Canvas. Notably, temporary chats are disabled for the model due to a technical issue, and could be fixed soon.