Glonce

Altman says OpenAI will have AGI by end of 2026, on his own definition

modelsThe Glonce editorsThe Decoder🔥76Glonce rating76Impact75Novelty65Relevance95Surprise55Players95Credibility70
Altman says OpenAI will have AGI by end of 2026, on his own definition

According to an extensive TIME report by journalist Alex Heath, who spent more than two weeks interviewing over 20 OpenAI executives, employees, investors and rivals, OpenAI's leadership is confident the company will have an internal system Sam Altman would call artificial general intelligence (AGI) by the end of 2026, even though they are "not quite yet" there. Chief Research Officer Mark Chen estimates OpenAI is "80% of the way" to AGI. Co-founder Greg Brockman believes people will look back on this period as the moment AGI emerged.

The confidence centers on Astra, OpenAI's upcoming model family, which the company announced last year as an automated AI research intern. Chief Scientist Jakub Pachocki told TIME that Astra already hits that internal benchmark: hand it an experimental idea and it can write the code in OpenAI's codebase, run the experiment and report back with results; give it a research paper and it can handle work that would take a human researcher about a week. Astra also enables "persistent agents," virtual coworkers that tackle tasks on their own over longer periods. Altman said during a customer preview that he expects Astra to be "the first model where the model actually invents new things in a way that matters," calling that "a very AGI-like thing." The underlying wager is that a system like this could trigger recursive self-improvement, where AI helps make itself better; researchers are split, with some seeing that capability as still far off and others reporting early signs it is already happening.

The AGI label itself is contested. OpenAI defines AGI as "highly autonomous systems that outperform humans at most economically valuable work," an economics-driven definition. Others argue AGI requires more, such as a robust understanding of the world, and that large language models are only one piece of that puzzle; whether systems built purely on language models can make genuine new discoveries and generalize, which the report frames as a basic requirement for AGI, remains an open debate.

TIME also reports OpenAI is building a "small handful" of hardware devices, according to Altman: something for a table, something for a pocket, and something to wear. The first is expected in early 2027, a small, puck-shaped device that senses its surroundings and talks to its owner via voice. Altman added that OpenAI will "definitely" build humanoid robots as well.

To pay for all of this, OpenAI is tightening its product focus. The key move is merging Codex, its coding agent, with ChatGPT in an internal project called "The Merge," which has produced ChatGPT Work, meant to put agentic AI in front of a mass audience by handling tasks rather than just answering questions. Thibault Sottiaux, the product lead running the project, told TIME that OpenAI is "close" to shipping something with "persistence and always-on execution," adding: "It's definitely going to feel like a new thing to people." OpenAI is also testing ads inside ChatGPT, which have performed well enough in early trials that leadership is increasing their volume. CFO Sarah Friar said the ad revenue could help cover costs for the 92 percent of ChatGPT consumer users who do not pay for a subscription. The company is also testing "Sponsored Agents," a format where clicking an ad drops the user into a brand-hosted AI experience.

Key facts

  • Sam Altman says OpenAI will have an internal system he would call AGI by the end of 2026, though "not quite yet" there; CRO Mark Chen puts progress at "80% of the way".
  • The claim rests on Astra, OpenAI's upcoming model family; chief scientist Jakub Pachocki says it already works as an automated research intern, handling work that would take a human researcher about a week when given a research paper.
  • OpenAI defines AGI narrowly, as systems that outperform humans at most economically valuable work; critics argue true AGI needs more, such as a robust understanding of the world.
  • OpenAI plans a puck-shaped AI hardware device for early 2027, the first of a "small handful" of planned devices, plus humanoid robots later.
  • To fund the push, OpenAI is merging Codex with ChatGPT into ChatGPT Work and testing in-app ads and "Sponsored Agents"; CFO Sarah Friar says ad revenue could help cover costs for the 92 percent of ChatGPT users who don't pay.

Why it matters

This is not a marketing claim from a press release. It comes from an investigative TIME report by Alex Heath, built on more than two weeks of interviews with over 20 OpenAI executives, employees, investors and rivals, and it puts named leaders (Altman, Chen, Brockman, Pachocki) on record with a specific timeline for an AGI-labeled internal system. Given how central the AGI milestone is to OpenAI's funding story, compute spending and public narrative, a concrete end-of-2026 target from the company's own leadership is a notable data point in the broader debate over how close the industry actually is.

Who it affects

AI researchers and rival labs, who now have a concrete internal claim to test against their own progress; OpenAI's investors and partners, who are betting on this trajectory; OpenAI's own staff building Astra and the merged Codex-ChatGPT product; and the ChatGPT user base, especially the 92 percent of consumer users who don't pay for a subscription and may see more ads as OpenAI leans on advertising to cover costs.

How to use it

There is no product to try yet from this claim itself. What is coming: a puck-shaped AI hardware device expected in early 2027, described as the first of several planned devices (something for a table, a pocket, and to wear); ChatGPT Work, merging Codex's coding-agent capabilities into ChatGPT for task execution rather than just answering questions, which OpenAI says is "close" to shipping; and ads plus a "Sponsored Agents" format already in early testing inside ChatGPT.

How solid is it

The reporting itself is solid: TIME's Alex Heath spent over two weeks interviewing more than 20 sources, including rivals and investors, not just OpenAI staff. But the AGI claims at the center of the story are self-reported by OpenAI's own leadership about a system, Astra, that has not shipped and whose capabilities cannot be checked from what is in the source. The AGI bar being claimed is also OpenAI's own definition, an economics-focused one that is more attainable than definitions requiring broader world understanding. The source does not say whether Altman's, Chen's, or Brockman's claims have been independently verified or tested by anyone outside OpenAI.

Risks and caveats

AGI has no single agreed definition, so an "AGI by end of 2026" claim is partly a matter of which definition is used, and OpenAI's own definition is notably narrower than some rivals'. Whether recursive self-improvement, AI helping to improve itself, is actually underway is disputed even among researchers cited in the report, some see it as far off, others report early signs. OpenAI also has a direct financial incentive to project confidence: the company needs to fund enormous costs, and is simultaneously ramping up ChatGPT ads and a new sponsored-agent ad format, which creates pressure to hype capabilities alongside monetizing its user base.

“I expect this will be the first model where the model actually invents new things in a way that matters. That's a very AGI-like thing.”

— Sam Altman, OpenAI CEO, quoted by TIME