> ## Content Index
> Fetch the complete content index at: https://www.tao.media/llms.txt
> Use this file to discover other available public pages before exploring further.

# Jensen Huang Says AGI Has Arrived After OpenAI’s GPT-6 Astra Launch
- URL: https://www.tao.media/jensen-huang-says-agi-has-arrived-after-openais-gpt-6-astra-launch/
- Published: 2026-09-07T13:06:50.000Z
- Updated: 2026-09-07T13:06:50.000Z
- Description: Nvidia’s founder and CEO says OpenAI’s newest model marks the arrival of artificial general intelligence, while another 400,000 GPUs are expected to come online as the company scales its Stargate infrastructure.
- Author: Bart Hillerich
- Tags: OpenAI, News, AI, Jensen Huang, Nvidia

Nvidia CEO [Jensen Huang](https://x.com/JensenHuang?ref=tao.media) said “AGI has arrived” after [OpenAI](https://openai.com/index/gpt-6-astra/?ref=tao.media) released GPT-6 Astra, a new frontier model the company says reaches state-of-the-art performance across advanced reasoning, coding, computer-use, science, health, and cybersecurity benchmarks.

Huang [made the declaration](https://x.com/JensenHuang/status/2096700264569090384?s=20&ref=tao.media) on September 6, three days after OpenAI introduced Astra as its “most intelligent and aligned model.” ([OpenAI](https://openai.com/index/gpt-6-astra/?ref=tao.media))

> “GPT-6 Astra, trained on \~100K+ NVIDIA Grace Blackwell NVLink72\. From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team. 400K GPUs coming online next.”

OpenAI President and co-founder Greg Brockman [echoed Huang’s assessment](https://x.com/gdb/status/2096721633876771094?s=20&ref=tao.media), though he left some ambiguity around exactly when the AGI threshold was crossed.

> “we're now moving into the AGI era (whether you view it as this model, the last one, or the next one), and could not do it without close partners”

## What OpenAI Announced With GPT-6 Astra

OpenAI launched GPT-6 Astra on September 3 as a new state-of-the-art model for use in computer use, browsing, software engineering, cybersecurity, science, and professional work. The company reported scores of 97.6% on FrontierMath Tier 4, 99.9% on ARC-AGI-3 and 100% on ExploitBench, alongside substantial gains in its ability to operate computers and complete multi-step professional workflows.

![](https://storage.ghost.io/c/78/0b/780ba906-b1a7-4bf0-873c-bdd5c32e5331/content/images/2026/09/image-7.png)

Astra can handle tasks such as filling out forms, updating records in a CRM, and organizing a calendar. The company said the model reached 59.3% on Agents’ Last Exam, a benchmark covering complex professional tasks in real software environments, and completed OSWorld 2.0 computer-use tasks with higher performance in about 47% less time than GPT-5.6 Sol.

Astra is also the first OpenAI model to reach the company’s Critical cybersecurity capability threshold. OpenAI said the model can discover previously unknown vulnerabilities and, without production safeguards, develop exploits against heavily protected systems, prompting additional monitoring and deployment restrictions. 

## The Implications of Huang's Claim

Huang’s post reflects Nvidia’s central role in the current AI buildout. Given GPT-6 Astra was trained on more than 100,000 Grace Blackwell NVLink72 GPUs, it makes the model as much of a capital-expenditure and physical infrastructure story as a software one. 

Frontier AI depends on large clusters of advanced accelerators, networking, energy, data-center construction, and cloud partnerships. The next phase of model development is being defined by the scale of physical infrastructure as much as by algorithmic design. 

Up next at Stargate Texas, backed by the OpenAI, Oracle, and SoftBank infrastructure push, 400,000 GPUs will come online. It's the kind of dedicated AI capacity that only a small group of companies can currently finance and operate.

Amidst this context, it is important to note that Nvidia is a major financial backer and investor in OpenAI, which is targeting an IPO in 2027, OpenAI CFO Sara Friar [has told employees](https://www.cnbc.com/2026/08/19/open-ai-ipo-timing-2027-friar.html?ref=tao.media).

Equally important to consider is the lack of a clear definition of what AGI actually is. For some, it means a system that can outperform humans across most economically useful cognitive work. For others, it means a model that can reason, plan, learn, and act across domains with minimal human direction.

As Mickey Hardy, CEO of Arcadia, [noted in his post on X](https://x.com/mickeyhardy/status/2096712464415265104?s=20&ref=tao.media):

> "This is genius by tech CEOs. Spend hundreds of billions of dollars racing toward AGI. Never establish a concrete definition of what AGI actually is. Eventually declare that you’ve achieved it. Then immediately introduce the next frontier, probably ASI, and raise another trillion dollars to chase that. The greatest part about racing toward a finish line nobody can define is that you can move it forever."

Is this Nvidia potentially propping up OpenAI before its IPO, and AGI remains a target off in the distance? Or is this the real deal?

We'll have to see once Astra gets into the hands of global users.

The company says the model is rolling out first to a limited set of organizations before expanding to ChatGPT Plus, Pro, Business, and Enterprise users, as well as developers through the OpenAI API, Microsoft Azure, and AWS Bedrock.