OpenAI unveiled its latest generation AI model, GPT-6 Astra, on Thursday, describing it as a state-of-the-art system for navigating computers and web browsers, writing software, and tackling complex mathematics. The company plans to roll out access to paying customers in the coming days, beginning with enterprise clients enrolled in its Daybreak early access program before extending availability to subscribers of ChatGPT Plus, Pro, and Business plans. Whether the model will reach free ChatGPT users was not clarified.
During a press briefing, OpenAI leadership expressed strong confidence in the new system. President and co-founder Greg Brockman told reporters he personally believes humanity has entered the era of artificial general intelligence—systems that are broadly smarter than humans across a range of tasks.
“It’s not unreasonable to feel that we are now in the AGI era,” Brockman said. “I think that if we fast-forward a couple of years, when we look back and say, ‘When was it really that AGI was created?’ I think it’s going to be about this time, and I think it might be about this model.”
The release comes at a pivotal juncture for OpenAI. The organization is racing to accelerate technological progress and grow revenue ahead of a planned initial public offering, all while attempting to maintain rigorous safety standards. Competition has intensified, particularly from rival Anthropic, which is also preparing for its own IPO.
In a accompanying blog post, OpenAI labeled GPT-6 Astra the “world’s best computer use model,” asserting that it can navigate computers and browsers on behalf of users far more effectively than prior iterations. The ability to reliably browse websites, complete forms, and manipulate software tools quickly and accurately has historically proved difficult for AI developers seeking to replace human workers.
According to OpenAI, testing demonstrated that GPT-6 Astra could book Department of Motor Vehicles appointments, search for job openings, and hunt for apartments more rapidly than the average person.
The company also claimed the model is its safest and most aligned release to date. Chief scientist Jakub Pachocki explained that OpenAI monitors the system’s chain-of-thought—its internal reasoning scratchpad—as a safety measure. He cautioned that ensuring advanced AI models do not produce unintended harm is growing increasingly difficult and could become a constraint on future progress.
“We kind of take this visibility for granted, and we are seeing that as model capabilities are increasing, monitorability is getting more challenging,” Pachocki said. “We think confidence in monitoring may constrain further development, because we would not accept degradation in our ability to monitor model alignment beyond a certain level. We would withhold scaling until we can regain enough confidence.”
Leave a Reply