Wharton's Ethan Mollick gets early GPT-6 access, says it can autonomously work for days

Ethan Mollick, a professor at Wharton, announced he has early access to GPT-6 and rated it as stunning. His current takeaway: GPT-6 is already good enough to autonomously do complex, meaningful work for him over several consecutive days. A longer write-up is promised but not yet published.

Confirmed

• Mollick said GPT-6 is good enough to autonomously carry out complex and meaningful work for him over multiple days, based on his hands-on trial
• He shared a fun example: a historically grounded simulation of the Library of Alexandria built with GPT-6
• He previewed a longer experience/review article, with details yet to come

Why it matters

• Multi-day autonomous completion of complex work is a core expectation for next-gen frontier models; if true, it marks a real gain in long-horizon autonomy
• Mollick is a well-known voice on practical AI use in education, and his first-hand impressions are often read as a reliable signal of model capability; before the full review, capability boundaries, version details, and availability remain undisclosed

2026-09-04 ~ 2026-09-04 · 5 related posts

• Ethan Mollick on early GPT-6 access: it autonomously does his work for days — emollick · 2026-09-04
• [source] GPT-6 Is Good Enough to Do Complex Work Autonomously for Days, Says Mollick — emollick · 2026-09-04
• Mollick: GPT-6 autonomously does meaningful work for days, built an open-source Alexandria Library simulation — emollick · 2026-09-04

2 near-duplicate retellings: emollick · hewarsaber