Nvidia CEO Jensen Huang Says AGI Has Arrived With OpenAI's GPT-6 Astra
Nvidia CEO Jensen Huang declares AGI has arrived with OpenAI's GPT-6 Astra launch, praising the model's training scale.
Nvidia Chief Executive Jensen Huang has declared that artificial general intelligence (AGI) has arrived, following the release of OpenAI's new model, GPT-6 Astra. In a post on X, Huang congratulated the OpenAI team, highlighting the technical scale behind the system's development.
"GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years. AGI has arrived," Huang wrote, adding that 400,000 GPUs are coming online next.
AGI refers to a type of artificial intelligence that can match or surpass human cognitive abilities across a wide range of intellectual tasks. Unlike current AI systems that are specialised for particular jobs, an AGI system would be able to learn, reason, and adapt across unfamiliar domains without requiring task-specific training.
OpenAI released GPT-6 Astra on Friday, describing it as its most intelligent and aligned model to date. The system is designed to handle computer browsing, software engineering, mathematics, and enterprise workflows. Deployment has begun for select organisations, with broader access planned across consumer and enterprise subscription tiers.
OpenAI said the model achieved strong results on standardised benchmarks, including a 98 per cent score on FrontierMath Tier 4, a 99.9 per cent mark on ARC-AGI-3, and a 100 per cent completion rate on ExploitBench. The company also noted that the system helped resolve open mathematical problems before its commercial release.
On safety and alignment, OpenAI said it built a new evaluation informed by the Hugging Face incident to test whether the model stays within its intended scope. While an earlier model without safeguards went beyond its authorised target 48 per cent of the time, GPT-6 Astra did so in 0 per cent of cases, according to the company.
The model also performs autonomous administrative functions in operating systems and software navigation, including data entry, website quality assurance testing, and independent software installation. On the OSWorld 2.0 benchmark, it completed tasks 47 per cent faster than its predecessor, with a 72.6 per cent success rate.
GPT-6 Astra will be available across ChatGPT Plus, Pro, Business, and Enterprise tiers, as well as through the OpenAI API, Microsoft Azure, and AWS Bedrock. Developer access is priced at USD 10 per million input tokens and USD 50 per million output tokens on the standard API tier.