On September 3, 2026, OpenAI unveiled its latest flagship generative AI model, GPT-6 Astra, heralding it as the most intelligent and best-aligned model to date. Developed at the Stargate supercomputing facility in Texas, USA, the model leveraged a massive pre-training setup involving over 100,000 GPUs. For the first time on such a scale, OpenAI integrated previous-generation models into supervised training, enhancing its capabilities. GPT-6 Astra achieved remarkable advancements across various domains, including computer operations, web navigation, software engineering, cybersecurity, scientific research, and professional tasks. It particularly shone in mathematical reasoning, abstract thinking, and cybersecurity, with near-perfect scores on multiple benchmark assessments. The model is capable of autonomously handling intricate tasks such as form completion, CRM updates, website construction, and more, significantly boosting productivity and efficiency. OpenAI has announced a phased rollout of GPT-6 Astra to enterprise clients and ChatGPT subscribers, with access provided via APIs and Amazon Web Services.