Report: OpenAI's Millennium Problem model is internal, "significantly more capable than GPT-6 Astra"

kimmonismus · x · 2026-09-09

Per Axios, the OpenAI model that solved the Navier–Stokes Millennium Prize problem is an internal model "significantly more capable than GPT-6 Astra." kimmonismus notes OpenAI having stronger internal models is no secret, but the "significantly" wording is bold. Rumors also point to a new "Fable" model with a new pretraining system, said to be far better than Astra. The news lands amid a credit controversy over the math result.

Related event: OpenAI says Navier-Stokes proof used a next-gen model far stronger than GPT-6 Astra(6 posts)→

Original post →

More from Models

Models channel →