Key Highlights
- Jensen Huang, CEO of Nvidia, proclaimed that artificial general intelligence has been achieved following OpenAI’s September 3 release of GPT-6 Astra
- The training process utilized over 100,000 Nvidia Grace Blackwell NVLink72 computing systems
- GPT-6 Astra achieved exceptional benchmark results: 98% on FrontierMath Tier 4, 99.9% on ARC-AGI-3, and a perfect 100% on ExploitBench
- Nvidia plans to bring an additional 400,000 GPUs online in the near future
- Skeptics remain: prominent AI researcher Gary Marcus disputes the AGI claim, highlighting the absence of a universally recognized AGI definition
On September 3, 2026, OpenAI launched its groundbreaking GPT-6 Astra model, positioning it as the company’s most advanced and ethically aligned artificial intelligence system to date. The announcement immediately captured widespread attention throughout the technology sector, particularly from Jensen Huang, Nvidia’s chief executive.
Taking to X on September 7, Huang made a striking proclamation: “AGI has arrived.” He extended his congratulations to OpenAI’s development team and highlighted Astra’s sophisticated training infrastructure as evidence of the remarkable progress in artificial intelligence capabilities.
The training regimen for GPT-6 Astra leveraged an impressive array of more than 100,000 Nvidia Grace Blackwell NVLink72 computing systems. Huang revealed plans to deploy an additional 400,000 Nvidia GPUs, indicating substantial upcoming investments in large-scale AI infrastructure development.
According to OpenAI, Astra represents their most sophisticated and ethically calibrated system to date. The organization claims the model demonstrates superior performance across multiple benchmark categories, including computer automation, software development, cybersecurity analysis, scientific research, and professional task execution.
Astra’s benchmark achievements include an impressive 98% score on FrontierMath Tier 4 and a near-perfect 99.9% on ARC-AGI-3. Most notably, it secured a flawless 100% rating on ExploitBench.
Astra’s Capabilities and Features
GPT-6 Astra is engineered to manage extended and intricate workflows that exceed the capabilities of its predecessors. The system can autonomously control computer systems on behalf of users with minimal human intervention, spanning domains such as software development and professional business operations.
The model’s cybersecurity functionality has generated significant interest within the industry. OpenAI reports that Astra has satisfied their internal “Critical” cybersecurity benchmark, demonstrating the ability to identify previously undiscovered software vulnerabilities and formulate exploitation strategies independently, without requiring detailed human guidance.
Given these powerful capabilities, OpenAI has implemented additional security measures and imposed restrictions on certain Astra features during the initial rollout phase.
AGI Claims Face Scrutiny
Huang’s bold statement hasn’t received universal acceptance. Greg Brockman, President of OpenAI, suggested that Astra might signal the dawn of the AGI age but carefully avoided declaring that the organization had definitively achieved artificial general intelligence.
Gary Marcus, a well-known skeptic in the AI field, challenged Huang’s assertion. Marcus emphasized that the AI community lacks a consensus definition of AGI, and there’s insufficient evidence demonstrating that existing systems have genuinely achieved this milestone.
The question of what truly constitutes AGI has remained a contentious topic among AI researchers for many years, and the Astra release has reignited this discussion in the public sphere.
Nvidia stock has climbed 23.7% year-to-date. Wall Street analysts on TipRanks have assigned Nvidia a Strong Buy consensus rating, derived from 29 Buy recommendations.
The consensus analyst price target for Nvidia stands at $325.23, suggesting potential upside of 41.2% from present price levels.
From Nvidia’s perspective, the massive scale of Astra’s training operation highlights robust and expanding demand for its artificial intelligence processors. The announcement regarding 400,000 additional GPUs entering service reinforces expectations for sustained chip demand throughout the remainder of 2026.
Get 3 Free Stock Ebooks
Discover top-performing stocks in AI, Crypto, and Technology with expert analysis.
- Top 10 AI Stocks - Leading AI companies
- Top 10 Crypto Stocks - Blockchain leaders
- Top 10 Tech Stocks - Tech giants
