Photo credit: www.techradar.com
OpenAI Introduces o3-mini Models to ChatGPTThe Announcement Follows AnticipationOpenAI Faces Heightened Rivalry from China’s AI Sector
OpenAI has officially rolled out its new o3-mini AI model to all users across various ChatGPT plans, including free accounts. This latest model showcases enhanced reasoning skills, particularly in fields such as mathematics, programming, and scientific inquiry.
According to OpenAI, the o3-mini release “advances the boundaries of what small models can achieve,” noting that it offers responses 24% quicker than the previous o1-mini model. External evaluations indicate that users prefer the answers generated by o3-mini over o1-mini 56% of the time, while also identifying 39% fewer errors in the new model’s outputs.
Similar to its predecessor, o1-mini, the o3-mini model will display its reasoning process alongside its answers, allowing users to follow the logical steps taken. Additionally, there is the option to integrate web searches with this reasoning, although this feature is still in the experimental phase.
This launch arrives in the wake of a competitive week in the AI landscape, during which China’s DeepSeek introduced high-speed, accurate models that quickly gained millions of users. In response, OpenAI aims to reclaim some of the attention and user engagement.
Give it a Try
The o3-mini is now accessible in ChatGPT and through the API. Users with pro accounts will enjoy unlimited access, while those subscribed to Plus and Team plans will have triple the rate limits compared to o1-mini. Free users can experiment with the o3-mini by selecting the Reason button located below the message input box.
Free users can access o3-mini by clicking the Reason button within the text entry section. OpenAI has not disclosed the exact usage limits, but they are expected to align with current restrictions placed on GPT-4o, likely allowing a limited number of queries per hour.
For subscribers, o3-mini can be chosen from the model selection menu on the top-left corner of the interface. Plus and Team plan members are allowed 150 daily queries, while Pro plan users enjoy unlimited access for a monthly fee of $200 (approximately £160 / AU$320).
In addition, subscribers will have the option to use an o3-mini-high model, which retains the same reasoning abilities but may take slightly longer to process responses, offering enhanced performance for those who are willing to wait a few extra moments.
OpenAI has also emphasized the comprehensive safety evaluations conducted on o3-mini before its launch, claiming that it “significantly surpasses” the GPT-4o model in evaluating unsafe usage and dodging jailbreak attempts.
Related Topics You Might Find Interesting
Source
www.techradar.com