OpenAI has released a new artificial intelligence model, o3-pro. The company calls it its most powerful model yet. It's an updated version of the o3 model, which specializes in logical thinking. Unlike traditional AI models, such models solve problems step by step, which makes them more effective in areas that require precise reasoning — such as mathematics, physics, and programming.
o3-pro is available to ChatGPT Pro and Team users, replacing the previous o1-pro model. Enterprise and Edu users will have access to the new model next week. In addition, the model is already available to developers via the API. Usage price: $20 per million inbound tokens and $80 per million outbound. One million inbound tokens is approximately equal to 750 thousand words.
According to OpenAI, o3-pro significantly outperformed the o3 model in all categories tested, especially in areas such as science, education, business, programming, and writing. Experts also highly rated its ability to follow instructions, clarity, completeness of answers, and accuracy.
o3-pro can use tools such as web search, file analysis, Python, image processing, and personalized memory. However, the model is slower than o1-pro and has several limitations: temporary chats in ChatGPT are currently unavailable, the model does not create images, and does not support the Canvas feature.
Despite these shortcomings, the o3-pro performed well in benchmarks. It beat Google's top model, Gemini 2.5 Pro, on the AIME 2024 math test, and outperformed Anthropic's Claude 4 Opus on the GPQA Diamond science test.
OpenAI positions o3-pro as the new standard for tasks requiring a high level of understanding and accuracy.