o1-pro

2025-03-19 · Publisher : OpenAI · Domain : Artificial Intelligence

OpenAI makes o1-pro available in the API: a version of o1 that devotes more compute to each response, reserved for the Responses API and billed at a very high price.

What changes

  • First "pro" version of a reasoning model offered to developers.
  • Available only in the Responses API (and Batch), to allow multi-turn interactions before answering.
  • Aimed at hard problems where consistency of answers matters more than latency.
  • Will be replaced in ChatGPT Pro by o3-pro on June 10, 2025.
  • Keeps the features of the o1 series: reasoning before answering (reasoning tokens supported) and image input.
  • Its pricing of 150 and 600 dollars per million tokens reserves it for occasional, high-value uses.

### Specifications - 200,000-token context, maximum output 100,000 tokens. - Text and image input, text output, reasoning tokens supported. - Knowledge cutoff: October 1, 2023. - API identifier: o1-pro (snapshot o1-pro-2025-03-19). ### Pricing Pricing per million tokens (USD), according to the documentation. | | Input | Output | | --- | --- | --- | | o1-pro | 150 | 600 | ### Benchmarks No official figure published at this date in the sources consulted (the API changelog entry cites no score). ### Key takeaways o1-pro remains reserved for cases where each response justifies a high unit cost. The price drop brought by o3-pro three months later (20 and 80 dollars) makes it quickly obsolete.

Links

Tags : LLM, OpenAI, o-series, Raisonnement, Pro