Key Takeaways
Fifteen months after the launch of GPT-4, the global tech community is buzzing with anticipation over the next generation of OpenAI’s language model series.
But while no official release date has been confirmed by OpenAI, a series of events and statements have fueled widespread speculation.
The excitement surrounding GPT-5 initially began to surge in November 2023, when OpenAI researchers hinted at ongoing projects that promise significant leaps in natural language understanding and generation at the company’s developer conference.
Six months later the company was still keeping hush about the rumoured project. But developments at Microsoft Build in May sent the reignited speculation.
According to AI writer Dan Shipper , who attended the event, during “every major presentation at Build, executives talked extensively about scaling laws.” From this, he concluded that “GPT-5 is coming, and it’s going to be amazing.”
Microsoft’s interest in scaling laws points to a company that believes the more resources it devotes to AI training, the more powerful the models it can build. And for now at least, the firm seems to expect it to continue scaling AI without suffering from diminishing returns.
Just 8 months separated the release of the first and second GPT models. But it was then 16 months before OpenAI released GPT-3 and nearly 3 years passed after that before GPT-4 was ready.
If the timeline of previous releases is anything to go by it could be years before GPT-5 hits the market. But earlier incarnations of the model were built in the pre-Microsoft era.
OpenAI today has significantly more resources, and the view of AI scaling espoused by Microsoft CEO Satya Nadella and others at Build 2024 suggests that building the next generation of AI is simply a matter of throwing enough compute at the problem.
With the GPT-5 rumor mill churning in overdrive, OpenAI CEO Sam Altman has remained tight-lipped about any specifics.
When asked about a timeline for the release of GPT-5 by Lex Fridman, he insisted “I don’t know,” adding that “we have a lot of other important things to release first.”
Of course, Altman has been through this before and made clear in the runup to releasing GPT-4 that OpenAI wouldn’t rush the launch:
“It’ll come out at some point when we are confident that we can do it safely and responsibly. I think in general we are going to release technology much more slowly than people would like. We’re going to sit on it for much longer than people would like.”
Meanwhile, OpenAI CTO Mira Murati has been slightly more candid about her expectations for future releases. Without naming the next GPT-model, in a recent interview Murati said that “in the next couple of years, we’re looking at PhD-level intelligence for specific tasks.”