OpenAI has launched GPT-6 Astra, its newest and most capable artificial-intelligence model, presenting it as a major advance in the evolution of ChatGPT from a conversational assistant into an increasingly autonomous digital worker.

Released on September 3, 2026, Astra is initially being made available to a limited group of organizations, with access for ChatGPT Plus, Pro, Business and Enterprise customers and API developers scheduled to expand in the following days. OpenAI describes Astra as its strongest model yet for computer use, software engineering, scientific research, cybersecurity, and complex professional work.

OpenAI President and co-founder Greg Brockman used unusually dramatic language to characterize the launch. Asked whether Astra could represent the arrival of artificial general intelligence, or AGI, Brockman told reporters, ‘I think it might be about this model,’ and concluded the briefing by saying, ‘Welcome to the AGI era.’

One of Astra’s most important advances is what OpenAI calls computer use. Instead of merely telling someone how to perform a digital task, Astra can potentially perform much of that task itself. It can navigate websites, complete forms, manipulate spreadsheets, update customer records, organize calendars, research information online and operate software applications.

Brockman said Astra can ‘zip through spreadsheets, fill out forms, and navigate across webpages often at superhuman speed.’ OpenAI says the model is also substantially faster than its GPT-5.6 predecessor on computer-use tasks.

For businesses, that could significantly expand the role of AI. Companies could use Astra to automate workflows that currently require employees to move repeatedly between databases, web pages, spreadsheets, and internal software. It can produce professional documents, presentations and spreadsheets while following company templates, conduct research, write and test software, analyze data, and perform multistep assignments with less constant human supervision.

Early business testing has already shown potential productivity gains: legal-technology company Legora reported that Astra reviewed 41 financial documents in minutes and detected all four deliberately inserted errors in one evaluation.

Individuals could experience a similar transition from AI assistant to AI agent. Rather than simply recommending apartments, jobs or government services, Astra can help search for them, compare options, and navigate portions of the application process. OpenAI has demonstrated applications ranging from apartment hunting and finding doctors to scheduling DMV appointments and helping prepare tax forms.

Those capabilities also create new risks. Astra is the first OpenAI model classified as reaching the company’s ‘Critical’ cybersecurity capability threshold, meaning that with appropriate tools it could discover previously unknown vulnerabilities and develop methods for exploiting sophisticated computer systems. To address these concerns, OpenAI says it has imposed stronger monitoring, security restrictions, and access controls around especially sensitive capabilities.

If Astra proves reliable and economical at scale, businesses may redesign jobs around supervising AI agents rather than manually completing routine digital tasks. Individuals could then increasingly delegate everyday computer work to AI. Whether or not historians ultimately regard GPT-6 Astra as the beginning of AGI, its launch represents an important step toward computers that do more than advise humans, they do more of the work themselves.