Jensen Huang declares: "AGI is here!" Highly praises OpenAI for releasing GPT-6 Astra

According to the Zhihui Finance APP, NVIDIA (NVDA.US) CEO Jensen Huang publicly stated after OpenAI released GPT-6 Astra that "Artificial General Intelligence (AGI) is already here." He pointed out that this milestone is thanks to the rapid evolution of AI models over the past four years, with the latest model trained using more than 100,000 Grace Blackwell NVLink72 chips.

"Next, 400,000 GPUs will be launched successively," Huang added, and offered his congratulations to OpenAI on the launch of Astra.

At 4:00 a.m. Beijing time on September 4, 2026, OpenAI officially released its new generation flagship model GPT-6 Astra, defining it as "the new generation of intelligence." Astra, meaning "stars" in Latin, is described by OpenAI as "currently the world's most intelligent and most aligned model," achieving advanced levels in computer operations, web browsing, software engineering, cybersecurity, scientific research, and professional work.

OpenAI President Greg Brockman stated at the release, "Welcome to the AGI era," but was noticeably more cautious in his wording: "We are now entering the AGI era (whether you view it as this model, the previous model, or the next one)."

OpenAI stated that Astra achieved a 98% score rate on the FrontierMath Tier 4 test, reaching a saturation point, and that this model has already helped resolve multiple longstanding open problems in mathematics. In addition, Astra scored 99.9% on the ARC-AGI-3 test and achieved a perfect 100% on the ExploitBench test.

OpenAI announced that GPT-6 Astra is now open to select institutions, and will gradually become available to all ChatGPT Plus, Pro, Business, and Enterprise users in the coming days. It will also be accessible via the OpenAI API, Microsoft Azure, and AWS Bedrock platforms.

From a business perspective, Jensen Huang's statement aligns closely with NVIDIA's interests. In the second quarter of the 2027 fiscal year ending July 2026, NVIDIA's data center business generated $89 billion in revenue, a 117% year-on-year increase, with total revenue reaching $96.2 billion. Huang stated bluntly on the earnings call: "Now, computation is revenue."

When a company's core business is selling chips to AI labs, every major model release is interpreted as validation of the "need to buy more GPUs." As early as March 2026, Jensen Huang said on the Lex Fridman podcast, "AGI is now," with his criteria being whether an AI can autonomously establish and manage a $1 billion-level tech company.

However, the relationship between NVIDIA and OpenAI is also somewhat delicate. In September 2025, the two parties announced a memorandum of understanding for AI infrastructure cooperation worth up to $100 billion, but in January 2026, The Wall Street Journal reported that negotiations stalled due to NVIDIA's internal doubts about OpenAI's business model. Against this backdrop, Huang's statement that "AGI is here" is seen by some observers as a PR move to maintain ties and boost market confidence.