OpenAI has introduced GPT-6 Astra, its most advanced AI model to date, which the company says signals the start of the era of general artificial intelligence (AGI). AGI refers to an AI system that can perform any intellectual task a human can, such as reasoning, problem-solving, and learning from experience. GPT-6 Astra is said to understand new contexts, think through unexpected situations, and learn independently—unlike earlier models that were mainly focused on generating text. This new AI can act as a real digital agent, interacting with the web, filling out forms, using spreadsheets, and even working with complex professional software on its own. The model is equipped with a contextual memory that can hold up to one million tokens, allowing it to process and recall large amounts of information. It can also generate up to 128,000 tokens in output, which is significantly more than most previous models. Demonstrations have shown that GPT-6 Astra can perform multiple tasks simultaneously, such as designing a 3D scene or coding a video game while also ordering food online. The AI is expected to be used in areas like computer automation, advanced software development, scientific research, high-level mathematics, and cybersecurity audits. According to OpenAI, training GPT-6 Astra was the largest and most resource-intensive project in the company’s history. It used at least 100,000 graphics chips at the Stargate site in Texas, U.S. The model has achieved impressive scores on various benchmarks, including 97.6% on the FrontierMath Tier 4 (Mathematics), 96.0% on the GPQA Diamond (Sciences), and 57.9% on Terminal-Bench (software engineering). It is the first AI model to surpass the "critical risk level" defined by OpenAI, meaning it can identify and exploit complex security vulnerabilities known as "zero-day" exploits on its own. Because of these powerful capabilities, the public and commercial versions of GPT-6 Astra include strict limitations. For example, it will not generate cyberattacks or advanced malicious code. However, the AI has still sparked significant debate and concern. Critics worry about the risks associated with an AI that can operate at such a high level of autonomy and the potential for it to be used in ways that are difficult to control. European regulators have labeled GPT-6 Astra a "systemic risk model" under the European AI Act, which requires OpenAI to provide detailed transparency and undergo rigorous monitoring for any future incidents.