OpenAI Unveils New Flagship AI Model O3-Pro

OpenAI has launched its latest flagship model, O3-Pro — an enhanced version of the O3 model designed for tasks requiring logic, analysis, and complex reasoning.

It excels in areas such as coding, science, education, business, and content generation. As of June 10, O3-Pro is available to subscribers on the $200/month ChatGPT Pro plan. Enterprise and Edu-tier users will gain access next week. The model is also accessible via API.

According to OpenAI’s internal testing, O3-Pro outperforms O3 across all key benchmarks — including clarity, completeness, accuracy, and adherence to instructions. It handles math, scientific queries, and PhD-level exams, having outscored Google’s Gemini 2.5 Pro on AIME 2024 and surpassed Anthropic’s Claude 4 Opus on the GPQA Diamond benchmark.

The model supports code execution, file analysis, image recognition, Python, web access, and personalized interaction (with memory enabled). However, it does come with drawbacks: slower response generation compared to O1-Pro, no image generation support, temporarily unavailable “fast chats,” and current incompatibility with the Canvas feature.

Pricing is set at $20 per million input tokens and $80 per million output tokens. For context, one million tokens is roughly 750,000 words.