Nvidia's Powerhouse Helps OpenAI Reach New Heights in AI Development
OpenAI has rolled out access to its latest AI model, GPT-6 Astra, to a limited number of organizations. The company plans to expand wider subscriber access soon.
The model was trained on over 100,000 Nvidia GPUs, according to Jensen Huang via X, who also stated that there are already plans to get 400,000 GPUs online in anticipation of whatever comes next.
Greg Brockman, co-founder and president of OpenAI, said this is the first time they've trained a model on more than 100,000 GPUs, leaving little doubt about how we ended up with a memory supply crisis.
Nvidia CEO Jensen Huang claimed that 'AGI has arrived' in an interview. AGI refers to Artificial General Intelligence, which broadly describes an AI model matching or surpassing human capability on complex cognitive tasks.
OpenAI's GPT-6 Astra is 'state-of-the-art' on several AI benchmarks, including FrontierMath Tier 4 and ARC-AGI 3. A trailer showcases its capabilities in creating presentations, eBay listings, and .STL files for 3D printers.