Model positioning
OpenAI o3-pro is a reasoning version of o3 that uses more compute to think for longer, designed for complex problems and tasks where answer reliability is important. It uses the same underlying model as o3. (help.openai.com)
Capabilities and uses
- Suitable for tasks requiring deep reasoning, such as mathematics, science, programming, education, business analysis, and writing assistance.
- In ChatGPT, the capabilities described in the release announcement include searching the web, analyzing files, reasoning about visual inputs, and using Python. Actual tool availability depends on the product environment.
- The API supports text and image inputs, text output, function calling, and structured outputs; the context window is 200,000 tokens, the maximum output is 100,000 tokens, and the knowledge cutoff date is June 1, 2024. (help.openai.com)
Limitations and status
- Complex requests may take several minutes; the official recommendation is to use background mode to avoid timeouts.
- The model documentation states that it is used through the Responses API; it does not support streaming, fine-tuning, audio or video input or output, or image generation.
- As of November 18, 2026, OpenAI had marked
o3-proand the snapshoto3-pro-2025-06-10as deprecated, with removal from the API scheduled for December 11, 2026. (developers.openai.com)
Release date
June 10, 2025. At launch, it was available to ChatGPT Pro users and through the API; product availability at launch does not represent current availability. (help.openai.com)
API pricing
The official model page lists text token prices as follows: 20 US dollars (USD) per 1 million input tokens and 80 US dollars (USD) per 1 million output tokens. Tool calls and other items may incur separate charges; batch processing or other discounted prices are not listed here. Date checked: November 18, 2026. (developers.openai.com)