OpenAI CEO Sam Altman admitted that the release of GPT-5 did not go as planned, as he acknowledged mistakes in how the upgrade was introduced to users.
He explained that the company is already taking those lessons into account while preparing its next model, GPT-6.
Altman spoke at a private dinner with reporters in San Francisco, as first reported by The Verge. He said the biggest error was switching ChatGPT’s default system from the “4o” model, which was popular for its natural tone, to GPT-5 without enough consideration for how disruptive that change would be.
Did you know?
Subscribe – We publish new crypto explainer videos every week!
What is Olympus DAO? (OHM Crypto Animated Explainer)
He said, “I think we totally screwed up some things on the rollout”. He added that pushing a sudden update to hundreds of millions of people at once was a mistake.
Altman said GPT-6 will arrive faster than the long wait between GPT-4 and GPT-5. However, progress depends on chip supply, which remains limited. He noted:
We have better models, and we just can’t offer them because we don’t have the capacity.
He explained that massive investment in data centers, possibly “trillions of dollars”, will be required to support future systems.
Altman also commented on the wider artificial intelligence (AI) industry. He said that investor enthusiasm for AI has likely gone too far, but at the same time, he believes the technology is one of the most important developments in years.
On August 15, the AI company Anthropic added a new option to its Claude models, Opus 4 and 4.1. What does the feature do? Read the full story.
