The OpenAI o3-mini, previewed in December 2024, has been launched by OpenAI as its “most cost-efficient mannequin” in its reasoning collection. The o3-mini’s power lies in science, math, and coding, and it is optimized for STEM reasoning.
OpenAI says that o3-mini gives a specialised different for tech domains that require velocity and precision, and with medium reasoning efforts, it matches the o1’s efficiency in science, math, and coding whereas delivering quicker responses. The corporate additionally mentioned that o3-mini was 24% quicker than o1-mini in A/B testing, with a mean response time of seven.7s in comparison with o1-mini’s 10.16s.
The o3-mini is OpenAI’s first small reasoning mannequin supporting extremely requested developer options, together with perform calling, developer messages, and Structured Outputs. Moreover, it should help streaming, and builders could have the choice to select from three reasoning efforts – low, medium, and excessive. o3-mini will even work with search to seek out up-to-date solutions with hyperlinks to related sources on the net.
The o3-mini is already out there to ChatGPT Plus, Crew, and Professional subscribers. It may be chosen from the mannequin picker, the place it should substitute the o1-mini. All paid customers could have the choice to decide on “o3-mini-high” within the mannequin picker if they’re prepared to commerce response time for greater intelligence.
ChatGPT Professional customers could have limitless entry to each o3-mini and o3-mini-high, whereas Plus and Crew customers will get the next fee restrict – up from 50 messages/day with o1-mini to 150 messages/day with the o3-mini.
ChatGPT’s free customers not having any subscription can attempt o3-mini by deciding on “Motive” within the message composer or by regenerating a response, making o3-mini OpenAI’s first reasoning mannequin that is out there to free customers in ChatGPT. It is also out there in Microsoft’s Azure OpenAI Service.