OpenAI Launches o3-mini as Its Most Affordable Reasoning Model

According to OpenAI, o3-mini gives a specialized substitute for tech fields that demand accuracy and speed. With moderate reasoning effort, it can match the o1’s performance in coding, arithmetic, and science while producing quicker answers. Additionally, according to the business, in A/B testing, o3-mini outperformed o1-mini by 24%, with an average response time of 7.7 seconds as opposed to 10.16 seconds for o1-mini.

OpenAI has launched the o3-mini, its “most cost-efficient model” in its reasoning series, which was first seen in December 2024. Because it is designed for STEM reasoning, the o3-mini excels in science, math, and coding.

Now, ChatGPT Plus, Team, and Pro subscribers can access the o3-mini. Picked from the model selector, it will take the place of the o1-mini. If one is ready to exchange response time for intelligence, all premium users will be able to select “o3-mini-high” in the model selection.

While Plus and Team users will have a greater rate limit—up from 50 messages per day with o1-mini to 150 messages per day with o3-mini—ChatGPT Pro users will have unrestricted access to both o3-mini and o3-mini-high.

OpenAI’s o3-mini, its first compact reasoning model, includes a number of much desired developer capabilities, including function calling, developer messaging, and structured outputs. Additionally, streaming will be available and developers will have a choice between three reasoning efforts: low, medium, and high. Additionally, o3-mini will interface with search to get up-to-date answers with links to relevant internet resources.

Function calling, developer messaging, and structured outputs are among the much sought-after developer features that OpenAI’s o3-mini, its first tiny reasoning model, supports. Moreover, developers will be able to select from three reasoning efforts—low, medium, and high—and streaming will be supported. To discover current responses with connections to pertinent online sources, o3-mini will also integrate with search.

With intelligence on par with OpenAI o1, OpenAI o3-mini offers enhanced efficiency and faster performance. O3-mini performs better on other math and factuality tests with medium thinking effort in addition to the STEM assessments mentioned above. O3-mini outperformed O1-mini in A/B testing, delivering responses 24% faster, with an average response time of 7.7 seconds as opposed to 10.16 seconds.