OpenAI's leadership believes AGI is within reach. According to an extensive TIME report by Alex Heath, the senior ranks agree on that, though they differ by degree.

  • Sam Altman — convinced the company will have an internal system he would call AGI by the end of the year, even though they are "not quite yet" there.
  • Mark Chen, chief research officer — estimates they are "80% of the way" there.
  • Greg Brockman, co-founder — believes people will look back on this period as the moment AGI emerged.

The report draws on more than two weeks of interviews with over 20 executives, employees, investors and rivals.

Astra: an automated research intern

At the center of this confidence is Astra, OpenAI's upcoming model family. The company announced this system 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 have taken a human researcher about a week.

Astra also enables "persistent agents": virtual coworkers that tackle tasks on their own over longer periods.

The claim itself is contested

Where Altman sees the biggest payoff is in generating new knowledge. During a customer preview he is reported to have said: "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."

The underlying wager is that such a system could kick off recursive self-improvement, where AI helps make itself better. Opinions are split: some researchers think that capability is still far off, others see early signs it is already happening.

There is a broader debate too, over whether AI built purely on language models can truly make new discoveries and generalize — which seems like a basic requirement for AGI.

Whose definition

This is the crux. The term has no precise definition, so the claim is partly a matter of interpretation.

OpenAI defines AGI as "highly autonomous systems that outperform humans at most economically valuable work," a heavily economics-driven reading. Others argue more is needed — a robust understanding of the world, for instance — and that language models are just one piece of the puzzle.

The distinction matters, because when a company sets both the goal and the measure of having reached it, an announcement that it has arrived becomes a declaration rather than a measurement.

Paying for it

Beyond the lofty ambitions, OpenAI has a much more mundane problem to solve: paying for all of it. The company's answer is tighter product focus. The key move is combining Codex, its coding agent, with ChatGPT in a project internally called "The Merge." That effort produced ChatGPT Work, which is supposed to put agentic AI in front of a mass audience.

There are hardware plans too. According to Altman the company is working on a "small handful" of devices: something that belongs on a table, something for your pocket and something you wear. The first will be a small, puck-shaped device that senses its surroundings and talks to its owner via voice, expected early in 2027. Altman added that OpenAI will "definitely" build humanoid robots as well.