Sam Altman Open AI CEO said that ‘GPT-4..5’, which was within the early reading of the launch, received greater than expected within the private test. He also said that there isn’t a expectation of ‘XAI’s Rock-3’.
Altman’s CEO posted on X (Twitter) on the 18th, “I attempted GPT-4.5, and there was loads more moments of ‘I felt AGI’ amongst luxury testers than expected!”
There is no such thing as a detailed explanation, but that is the last non-discipline foundation model of open AI, which can be released ahead of the ‘GPT-5’ integrated with O3. It is understood that despite two pre -training last yr, it has not recorded a transparent performance improvement.
Nonetheless, the tweet of Altman’s CEO on the day is that the response is sort of good within the private test. Specifically, if you happen to mention artificial intelligence (AGI), evidently the performance is sort of higher than the present GPT-4O.
On this regard, the most recent version of GPT-4O (CHATGPT-4O-LATEST-20250129), which was released by Open AI the day before, also responded that some users experienced AGI moments. Nonetheless, it is just not clear whether this model is a GPT-4.5 test version.
GPT-4.5, called ‘Orion’, is understood to have already developed. In accordance with Bloomberg, Altman’s CEO foretelled his employees that “it’ll be released in just a few weeks.” After a security test, it’ll be released immediately.
Meanwhile, Altman also mentioned that Rock-3 in X on this present day.
He identified that a user identified that he watched that rock-3 within the comments, “I’m not good (that willn’t be very nice …).” That is written just a few hours ago.
By Dae -jun Lim, reporter ydj@aitimes.com
