OpenAI Launches GPT-6 Astra, Calling It a Major Milestone Toward Artificial General Intelligence

en.Wedoany.com Reported - On September 3, local time, OpenAI released its next-generation AI model, GPT-6 Astra, and began offering it to enterprise customers in the Trusted Access Program. OpenAI plans to make the model available via API and through ChatGPT Plus, Pro, Business, and Enterprise plans in the coming days. The company positions GPT-6 Astra as its most capable model to date, with a focus on enhancing performance in complex tasks such as computer operation, web browsing, software engineering, cybersecurity, scientific research, and professional work.

GPT-6 Astra can execute multi-step workflows across code environments, browsers, and professional software, and can directly complete computer operation tasks. Model documentation released by OpenAI shows that Astra supports a context window of up to 1.05 million tokens and a maximum output of 128,000 tokens, offering multiple reasoning effort levels ranging from low to max. Compared with previous models, its capabilities have extended further from single-turn question answering and code generation to long-duration, end-to-end task execution.

Cybersecurity capability is one of the core changes in this release. OpenAI confirmed that GPT-6 Astra has reached the "Critical" cybersecurity capability threshold in the company's Preparedness Framework, making it the first OpenAI model to achieve this level. When equipped with the appropriate tools and access permissions, the model can identify previously unknown security vulnerabilities without step-by-step human guidance and develop exploit methods against multiple highly protected systems.

In response to this capability level, OpenAI has simultaneously tightened model deployment restrictions, including enhanced internal environment isolation, encryption of model checkpoints, monitoring of full task trajectories, and the addition of blocking alignment evaluations before internal use. The company also launched the "Daybreak for Frontline Defenders" program, committing $1 billion to provide cutting-edge cybersecurity models and tools to defenders of critical infrastructure through subsidized access, training, technical support, and collaborative projects.

OpenAI President Greg Brockman said during the model launch that, looking back at this technological phase in the future, GPT-6 Astra may be regarded as a node in the early stages of artificial general intelligence. OpenAI did not directly define GPT-6 Astra as having achieved AGI in this release, but instead continues to treat it as a significant phase in capability advancement and autonomous execution of complex tasks by the model.