OpenAI has launched a new «reasoning model» o3-mini in the free version of ChatGPT (but with limitations)

OpenAI has finally launched the latest version of its reasoning models o3-mini in API services and ChatGPT subscriptions(including in the free version).

According to the company, o3-mini duplicates or even outperforms its predecessor o1 in a number of coding and reasoning tasks, but requires less cost and time (24% faster than o1-mini). Also, similar to o1-mini, the new model will show how it arrived at the answer, not just provide it.

To use o3-mini in the free version of ChatGPT, you need to press the «Reason» button. However, there are limitations — up to 10 queries (they are restored in a few hours after use). Plus and Team subscription holders have a limit of 150 requests, while ChatGPT Pro offers unlimited access.

Like DeepSeek, o3-mini also has a button to search for answers on the Internet and can provide links.

For paid users, the o3-mini model is offered in three levels of reasoning: low, medium, and high (the name corresponds to both the level of intelligence and the speed of the model’s answers).

Microsoft announced that o3-mini also available for the Microsoft Azure OpenAI Service, GitHub Copilot, and GitHub models.