OpenAI chief Sam Altman is currently touring the world. Following Japan, Korea, and India, Altman visited Berlin to talk about AI and the future. In a panel discussion on “The Age of AI” at TU Berlin, Sam Altman challenged the audience: “How many of you think you’re still going to be smarter than GPT-5?“
He then added, “I don’t think I’m going to be smarter than GPT-5. And I don’t feel sad about it because I think it just means that we’ll be able to use it to do incredible things. And you know like we want more science to get done. We want more, we want to enable researchers to do things they couldn’t do before. This is the history of, this is like the long history of humanity.”
Altman’s comment suggests that OpenAI is nearly done training the next-generation GPT-5 model, and post-training evaluations may be hinting towards broader intelligence.
That said, in a recent Reddit AMA, Altman said that the company doesn’t have a release timeline for GPT-5 yet. But in recent weeks, Altman has been speaking more openly about GPT-5 in interviews and press briefings.
Whatever the case, currently, OpenAI is focused on its o-series reasoning models. The o3 model is a breakthrough as it cracked the ARC-AGI benchmark, although, at a significantly higher cost. Not to mention, the Deep Research agent which is powered by the o3 model achieved 26.6% on Humanity’s Last Exam with web search and Python tools. Other competing models have only got around 9.4% without web search.
What is interesting is that during the tour in Japan, Sam Altman said that the company wants to integrate the GPT-series and o-series models to create one integrated model, the AGI (Artificial General Intelligence). I think, at some point, the GPT series will converge with the o-series to create an inference-scaled model that is far more capable and intelligent.
Source link
-
-
-
-
-
-
-
-
