OpenAI Unveils o3-Pro, Its Most Advanced Reasoning AI Model
OpenAI introduces o3-pro, a powerful upgrade to its step-by-step reasoning AI that outperforms competitors in math and science benchmarks. The new model is rolling out to ChatGPT subscribers and developers, though it comes with longer response times and a few missing features.
OpenAI releases o3-pro, a highly capable upgrade to its existing o3 reasoning model that tackles complex problems step by step. This new model immediately replaces o1-pro for ChatGPT Pro and Team users, with Enterprise and Edu access following next week. Developers can also integrate o3-pro through the API, where it costs $20 per million input tokens and $80 per million output tokens.
In internal expert evaluations, reviewers consistently favor o3-pro over the standard o3 model in categories like science, programming, and business. The AI demonstrates superior performance on major benchmarks, notably beating Google's Gemini 2.5 Pro on a math evaluation and outscoring Anthropic's Claude 4 Opus on a PhD-level science test. Additionally, o3-pro utilizes built-in tools to search the web, analyze files, run Python code, and process visual inputs.
Despite its impressive capabilities, o3-pro presents a few drawbacks for users to consider. The model takes longer to generate responses compared to o1-pro, and OpenAI temporarily disables temporary chats due to a technical issue. Furthermore, o3-pro currently lacks the ability to generate images and does not support OpenAI's Canvas workspace feature.