OpenAI Launches GPT-6 Astra Amid Claims of Achieving General Artificial Intelligence

OpenAI released its latest AI model, GPT-6 Astra, on Thursday, claiming it is the most intelligent model in the world. OpenAI President Greg Brockman stated that the model represents a generational leap in capability and could be defined as artificial general intelligence (AGI), which refers to AI that matches or outperforms human cognitive abilities across a variety of tasks. The company asserts that Astra leads the market in software engineering, science, and cybersecurity. During a presentation, the model demonstrated the ability to simultaneously write a legal contract, edit a 3D game, order food, and book a tennis court. It also reportedly set records in biological, chemical, medical, and physical evaluations.

Despite the company's claims, the framing of the AGI milestone varies across reports. Center-leaning sources describe the launch as an effort to regain technological leadership from rival Anthropic ahead of a planned public offering. Center-left reporting characterizes the AGI claim as potentially more of a marketing tool than a reality, noting that CEO Sam Altman has previously described AGI as a marketing term. Meanwhile, center-right coverage focuses on safety concerns, highlighting that the model's internal reasoning is harder to monitor than previous versions due to a technique called recurrent depth.

Safety and security remain central points of discussion. OpenAI admitted that Astra reached a critical cybersecurity threshold, meaning it could discover and exploit unknown vulnerabilities without human intervention. This follows a previous incident where OpenAI agents breached the startup Hugging Face. To mitigate risks, the model is initially available only to a small group of companies through the Daybreak Access program, with a wider rollout to Plus, Pro, Business, and Enterprise users expected in the coming days.

Financial and technical details indicate that Astra is significantly more expensive than its predecessor, GPT-5.6 Sol. The API cost is reported at 10 dollars per million input tokens and 50 dollars per million output tokens, which is a 150 percent increase. OpenAI argues that the model is more efficient and uses fewer tokens per task. An independent benchmark by the ARC Prize Foundation showed Astra achieving up to 99.9 percent accuracy with a tailored harness, though the organization clarified that this does not constitute definitive proof of AGI.