Tech
OpenAI Unveils o3-mini Model: Sam Altman Describes It as ‘Smart and Fast’
OpenAI Unveils o3-mini Model: Sam Altman Describes It as ‘Smart and Fast’
The company’s most recent AI reasoning model, o3-mini, was introduced by OpenAI CEO Sam Altman. Altman had hinted at the new o3-mini model’s impending shipment two weeks prior, while DeepSeek, a Chinese AI firm, was enjoying its success. The o3-mini model is now available within ChatGPT and its API services. Additionally, the business has released a version of ChatGPT with rate constraints for the first time to its free customers.
In his X post, Altman characterized the new model as quick and intelligent. According to him, the model is accessible to free-tier users by clicking the reason button, and it can search the web and display its thought processes. Users of ChatGPT Plus can choose the “o3-mini high,” a version of the model that thinks more deeply and provides better responses.
The o3-mini model was shown last month as a part of OpenAI’s 12-day “Ship-mas.” The model was created to outperform the existing reasoning model and equal O1’s performance in domains like science, math, and coding.
The newest and most affordable model in our reasoning series, OpenAI o3-mini, is being made accessible today via ChatGPT and the API. This robust and quick model, which was first seen in December 2024, pushes the limits of what small models can accomplish. It offers remarkable STEM capabilities, with a focus on science, math, and coding, while retaining the low cost and lower latency of the OpenAI o1-mini, the company said on its official website on Friday.
What is OpenAI’s o3-mini model?
Function calls, developer messages, and organized outputs are some of the characteristics of the new o3-mini, a tiny reasoning model created for production. The model can handle difficult jobs or even prioritize speed because it enables streaming and has three reasoning options: low, medium, and high.
The model, which triples the rate restriction to 150 messages per day, is being made available to ChatGPT Plus, Team, and Pro customers. The new model lacks eyesight in comparison to O1, but it incorporates internet search to provide current responses. Pro users will have unrestricted access to o3-mini and o3-mini high, while free users will be able to view the model for the first time.
O3-mini performs noticeably better than O1-mini when compared to benchmark performance, particularly in STEM subjects. O3-mini outperforms earlier models in mathematics, with an accuracy of 83.6 percent on the AIME 2024 challenge. Furthermore, with an average reaction time of 7.7 seconds as opposed to 10.16 seconds, o3-mini provides replies 24% faster than o1-mini.
