OpenAI is preparing to launch three new AI models: ‘o3’, ‘o4-mini’ and ‘o4-mini-high’.
This is indicated by code from the ChatGPT web app. The models are expected to offer better reasoning capabilities than the current versions. The new models are in line with OpenAI’s existing O series. ChatGPT currently has access to o1, o3-mini and o3-mini-high, which are known as reasoning models. The platform also offers GPT-4o and GPT-4.5.
OpenAI CEO confirms launch
Sam Altman, the CEO of OpenAI, confirmed last week on the social network X that the new models are coming. “We are going to release o3 and o4-mini after all, probably in a couple of weeks, and then do GPT-5 in a few months,,” he wrote in his message.
According to Altman, this decision has several reasons, but the most important is that OpenAI expects GPT-5 to become much better than originally thought. He indicated that integrating all the functions proved more difficult than anticipated and that the company wants to ensure sufficient capacity to meet the expected demand.
Improved reasoning capabilities
The full-fledged o3 model is expected to be a significant improvement over the current mini versions (o3-mini and o3-mini-high). These smaller models in the o-series are known for their improved performance and faster response times, but are still limited compared to what a full o3 model could offer.
The new o4-mini and o4-mini-high models will, in turn, probably be comparable to the existing mini versions but with improved reasoning capacities.
Tip: Grok swears and screams as uncensored ChatGPT alternative