OpenAI released o3-pro on Tuesday, a new AI model based on its o3 reasoning model. The company asserts o3-pro is its most capable model to date, with availability starting for ChatGPT Pro and Team users.
The new o3-pro, an enhanced version of the o3 reasoning model launched earlier this year, utilizes a step-by-step problem-solving approach, improving reliability in areas such as physics, math, and coding, unlike conventional AI models.
ChatGPT Pro and Team users gained access to o3-pro on Tuesday; it replaced the o1-pro model. Enterprise and Edu users will gain access the following week, according to OpenAI. The o3-pro model is also available in OpenAI’s developer API as of Tuesday afternoon.
The pricing for o3-pro in the API is $20 per million input tokens and $80 per million output tokens. A million input tokens is approximately 750,000 words.
OpenAI stated in a changelog that, in expert evaluations, reviewers preferred o3-pro over o3 across all tested categories, particularly in science, education, programming, business, and writing. The company added, “Reviewers also rated o3-pro consistently higher for clarity, comprehensiveness, instruction-following, and accuracy.”
According to OpenAI, o3-pro has access to tools that allow it to search the web, analyze files, reason about visual inputs, use Python, and personalize responses using memory. OpenAI notes a longer response time compared to o1-pro as a drawback.
Temporary chats with o3-pro in ChatGPT are currently disabled due to a “technical issue.” The model cannot generate images, and OpenAI’s Canvas workspace feature is not supported.
Internal testing by OpenAI showed that o3-pro achieved high scores on AI benchmarks. It outperformed Google’s Gemini 2.5 Pro on AIME 2024, a math skills test, and surpassed Anthropic’s Claude 4 Opus on GPQA Diamond, which assesses PhD-level science knowledge.
