Jensen Huang Says AGI Arrived With OpenAI GPT-6 Astra

For years, artificial general intelligence – the hypothetical point at which AI can perform a broad range of intellectual tasks at roughly human or better levels – has lived somewhere between a research goal and a science-fiction plot. Now Nvidia CEO Jensen Huang has made a dramatic declaration that this future has arrived, triggering intense debate across the technology sector.

Huang declared “AGI has arrived” on social media platform X after OpenAI launched GPT-6 Astra, its latest and most capable model. The chipmaker executive congratulated the OpenAI team while highlighting the massive computing infrastructure behind the new system, which was trained using more than 100,000 Nvidia Grace Blackwell NVLink72 systems, with another 400,000 Nvidia GPUs coming online.

OpenAI itself has been more measured in its claims. President Greg Brockman said Astra could mark the beginning of the AGI era, while the company has described the model as a major leap in capability. However, no universally accepted test officially certifies a model as AGI, leaving the industry without a clear benchmark for such ambitious claims.

What Makes GPT-6 Astra Different From Previous Models

Astra is not simply designed to answer questions faster – OpenAI says it is built to do things. The model can operate a computer, fill online forms, update customer records, conduct web research, and analyze data. It also creates websites, works with specialized scientific software, and produces documents, spreadsheets, and presentations.

The company describes Astra as its most intelligent and aligned model yet, with improvements spanning computer use, software engineering, cybersecurity, science, and professional work. This distinction matters because the AI industry’s center of gravity is gradually moving from “ask a chatbot something” to “give an AI a task and let it work through the steps.” Reuters described Astra as part of this broader shift toward agentic AI – systems capable of carrying out complex tasks with little human intervention.

OpenAI reported that Astra scored 98% on FrontierMath Tier 4, 99.9% on ARC-AGI-3, and 100% on ExploitBench. These benchmark results represent significant performance improvements over previous models, though critics question whether high scores on specific tests equate to genuine general intelligence.

Why Huang’s Declaration Carries Unusual Weight

Huang made his declaration while highlighting just how much computing power sits underneath the model. That makes his statement particularly notable because Nvidia supplies much of the computing infrastructure behind the current AI boom. Huang is therefore not simply commenting from the sidelines; his company’s hardware is part of the machinery powering these frontier models.

“From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team,” Huang wrote on X.

This is not the first time Huang has made such a claim. In March 2025, he said he believed AGI had already arrived during an appearance on the Lex Fridman Podcast, stating “I think we’ve achieved AGI.” His consistent position on this topic suggests a genuine conviction rather than mere marketing rhetoric, though it also raises questions about his definition of artificial general intelligence.

Prominent AI Expert Disputes Huang’s AGI Claim

Gary Marcus, an NYU professor and longtime AI skeptic, was far less convinced by Huang’s declaration. Marcus quickly pushed back against the claim, arguing that it lacks both evidence and clear definitions.

“Sad to see Jensen claim that AGI has arrived, with no evidence and no definitions,” Marcus wrote.

He argued that Astra still falls short under conventional definitions of AGI and warned that declaring victory without first agreeing on what AGI means “simply muddies the waters.” Marcus urged Huang to consider established frameworks for evaluating AGI claims before making such sweeping pronouncements.

According to Stanford’s Institute for Human-Centered AI, AGI refers to AI that can learn, reason, and apply knowledge across a broad range of domains at human-level competence or beyond. The institute’s definition highlights the need for flexible, general-purpose intelligence rather than narrow, task-specific performance, no matter how impressive.

The AGI Debate Reflects Broader Industry Tensions

The disagreement between Huang and Marcus reflects broader tensions within the AI community about how to measure progress toward true general intelligence. While some industry leaders see recent breakthroughs as evidence that AGI has arrived or is imminent, many researchers remain cautious about making premature declarations.

The debate is not merely academic. Declaring that AGI has arrived carries significant implications for regulation, investment, public perception, and the trajectory of AI development itself. If the technology industry cannot agree on what constitutes AGI, it becomes nearly impossible to have meaningful conversations about whether it has been achieved.

OpenAI president Greg Brockman shared Huang’s post and said the company is “now moving into the AGI era” while acknowledging uncertainty about whether this model, the last one, or the next one should be considered the true threshold. His hedged language suggests even those closest to the technology recognize the difficulty of pinpointing AGI’s arrival.

The shift toward agentic AI that Astra represents may indeed constitute a meaningful step change in AI capabilities. Systems that can autonomously complete complex, multi-step tasks across different domains demonstrate abilities that previous generations of AI could not match. Whether these abilities rise to the level of artificial general intelligence remains a question that may not have a clear answer without broader consensus on definitions and benchmarks.