Jensen Huang Says “AGI Has Arrived” as GPT-6 Astra Pushes AI Toward Delegable Workflows

TL;DR

On September 6, 2026, NVIDIA founder and CEO Jensen Huang congratulated OpenAI on the launch of GPT-6 Astra on X and wrote, “AGI has arrived.” This is an industry leader’s judgment about the milestone, not proof that AGI has a universally accepted definition or independent certification. The verifiable shift is that OpenAI is pushing GPT-6 Astra toward highly autonomous work across computer use, software engineering, cybersecurity, science, and professional tasks.

What happened

• Jensen Huang congratulated the OpenAI team on GPT-6 Astra and posted, “AGI has arrived.”
• OpenAI says Astra focuses on computer use, browsing, software engineering, cybersecurity, science, and professional work, with availability expanding from limited organizations to ChatGPT and API users.
• OpenAI also published a research update saying it had reached its goal of an automated research intern: under human direction, the system can handle well-defined research tasks that would take a skilled researcher several days.

Why it matters

The important change is not only the AGI label. Frontier model competition is moving from chat answers toward delegable, executable, and verifiable workflows. Developers should watch whether a model can reliably use computers and browsers, complete multi-step tasks, stay within authorization boundaries, and reduce the cost of real work.

Implications for developers and AI business opportunities

1. Agent workflows are becoming a major opportunity: browser operations, code changes, research, spreadsheets, and document work can become repeatable services.
2. Evaluation and verification will become more valuable. As vendors make stronger AGI claims, users need real-task tests, cost comparisons, failure cases, and safety boundaries.
3. This news should lead to tutorials on computer use, agent task design, permissions, and cost control. A case study should wait until there is verifiable revenue or customer impact.

Important caveat

AGI does not have one universally accepted test standard. This article reports “AGI has arrived” as Jensen Huang’s public view, not as an independently verified scientific conclusion. Performance figures on OpenAI’s page are vendor-reported results and should be checked against real tasks and third-party evaluations.

Sources

• Jensen Huang’s original X post
• OpenAI: GPT-6 Astra
• OpenAI: Research acceleration