Latest
'Alien: Isolation 2': Facing a ruthless and evolved Xenomorph in our Gamescom hands-on previewVance pushes back on reports that US strikes hit wedding in Iran but says ‘sometimes things happen’ – liveUS senator calls on the NSA to give guidance for use of VPNsMissouri House candidate charged in alleged plot to plant cocaine on opponent‘I wouldn’t call it a war,’ Vance says after Iran hostilities flareFrancis Fukuyama on changing your mind -The Cybercab is Tesla’s ‘fork in the road’ moment'Alien: Isolation 2': Facing a ruthless and evolved Xenomorph in our Gamescom hands-on previewVance pushes back on reports that US strikes hit wedding in Iran but says ‘sometimes things happen’ – liveUS senator calls on the NSA to give guidance for use of VPNsMissouri House candidate charged in alleged plot to plant cocaine on opponent‘I wouldn’t call it a war,’ Vance says after Iran hostilities flareFrancis Fukuyama on changing your mind -The Cybercab is Tesla’s ‘fork in the road’ moment
← Signal: Tech & AI

OpenAI Launches GPT-6 Astra, Positioning It as Potential AGI Milestone

Developing1 source · Sep 3, 2026

OpenAI released GPT-6 Astra, claiming it is state-of-the-art at computer navigation and software tasks, with executives suggesting it may mark the beginning of the artificial general intelligence era.

OpenAI Launches GPT-6 Astra, Positioning It as Potential AGI Milestone
Image via Wired

What happened

OpenAI announced GPT-6 Astra on Thursday, positioning it as the company's most capable model for navigating computers, browsing the web, writing software, and solving complex math problems. The model will roll out first to enterprise clients in OpenAI's Daybreak early access program, followed by ChatGPT Plus, Pro, and Business subscribers; OpenAI did not confirm whether free users will have access. OpenAI claims GPT-6 Astra demonstrated superior performance in real-world tasks, completing DMV appointments, job searches, and apartment hunts faster than average humans. President Greg Brockman stated he believes the world is now in the AGI era and suggested this model may be identified retrospectively as the point where AGI emerged. OpenAI also claims GPT-6 Astra is the company's safest and most aligned model, with Chief Scientist Jakub Pachocki noting the company monitors the model's internal reasoning processes to ensure safe deployment.

Context

The release occurs as OpenAI prepares for a planned initial public offering and faces intensifying competition from Anthropic, which is also pursuing an IPO. The timing reflects OpenAI's dual imperative to advance capabilities and maintain market position while managing safety concerns. Pachocki's statement that monitoring challenges may constrain future scaling signals that OpenAI views transparency into model reasoning as a fundamental constraint on further development—the company has stated it will withhold scaling improvements if monitoring confidence falls below acceptable thresholds. The broader context is the ongoing industry struggle to build AI systems that can reliably manipulate software interfaces and tools, a capability that has historically been difficult for AI models to master with sufficient reliability and speed.

What's disputed

OpenAI's characterization of GPT-6 Astra as marking the AGI era is presented as the personal belief of executives rather than a consensus view or objective assessment. The claim that GPT-6 Astra outperforms average humans on specific tasks is based on OpenAI's own tests and has not been independently verified.