For several months we have been talking insistently about GPT-5the next large language model (LLM) Of OpenAI.
Despite various rumors and a release that, from time to time, always seems imminent, for the moment the company has not yet made this important move.
This time, however, it could be the right time. As reported by the site Business Insider, in contact with two anonymous sources working in the sector, GPT-5 could see the light next summer. According to the two figures, OpenAI is already testing a first demo of the new AI on some corporate clients selected.
In this sense, those who were able to get their hands on the demo seem to have been very impressed by how effective it proved to be, with concrete improvements compared to GPT-4.5.
The release of GPT-5? Much will depend on safety testing
During the demo presentations, apparently, some previously unreleased functions of the highly anticipated LLM were shown. We are talking about an AI agent capable of executing tasks that are completely autonomous from the assistance of human beings.
Beyond the tests, however, OpenAI is certainly not sitting idle. In fact, the company of Sam Altman, is still powering GPT-5. Not only that: what makes the release likely only in a few months are the safety tests which, as expected, will be of fundamental importance for the final green light. If security problems arise, it is not unlikely that the debut of GPT-5 could be postponed further.
As already hinted, the new LLM will be mainly aimed at corporate customers, with the concrete possibility that OpenAI will “imitate” Gemini with different versions of the model, with different purposes and for different customer targets.