.png){kind=logo}
Nvidia CEO Jensen Huang declared on X on September 6 that artificial general intelligence has arrived, crediting OpenAI's new GPT-6 Astra model, which was trained on more than 100,000 Nvidia Grace Blackwell NVLink72 systems, in a post congratulating the OpenAI team.
Jensen Huang posted on X, writing that the journey from ChatGPT to o1 to Astra took just four years and that AGI has now arrived. He credited OpenAI's team directly and pointed out that Astra was trained using more than 100,000 Nvidia Grace Blackwell NVLink72 systems, adding that another 400,000 GPUs would soon come online. Huang had first mentioned a figure of 300,000 systems before deleting that post and replacing it with the smaller number.
OpenAI launched GPT-6 Astra on September 3, describing it as its most capable and aligned model so far. The company said Astra performs at a state-of-the-art level on tests like FrontierMath Tier 4, ARC-AGI-3, and TerminalBench-4.0, and also scored 100 percent on ExploitBench. OpenAI built the model to handle a wide range of work, including computer use, software engineering, cybersecurity, and scientific research.
Also read: Can the NVIDIA GTX 1650 Still Handle 1080p Gaming in 2026?
OpenAI President Greg Brockman had already framed the launch as the beginning of what he called the AGI era, and Huang's comments added weight to that claim coming from outside the company.
AGI does not have one fixed definition across the industry. OpenAI itself describes it as a highly autonomous system that can outperform humans at most economically valuable work. Many researchers argue the term gets used loosely, since Astra still needs task-specific guidance in several areas and has not been independently verified as matching human-level reasoning across every domain.
Huang's declaration has already drawn pushback from AI researchers who see it as more of a marketing statement than a technical milestone for Nvidia, especially given Nvidia's own financial stake in the success of large AI training runs. The debate around what actually counts as AGI is likely to continue as more labs release models with similar claims.
