The company says training began on 28 August, and even its own mathematicians were surprised by the speed of progress. OpenAI has excluded the pace of its research from the group's advisory role, meaning the group can advise on sharing results but not on how fast the model is developed. The advisory group is led by Fields Medalist Timothy Gowers and includes other prominent mathematicians, though Gowers did not sign an open letter warning that AI-generated solutions could undermine conceptual understanding.

OpenAI's chief scientist previously said the team deliberately chose not to optimise for math and was focused on recursive self-improvement instead. The company has admitted it only took on the Navier-Stokes Millennium Problem after hearing rumours that another team, partly made up of Anthropic researchers, had already cracked it. Skeptics suggest the publicity around solving famous problems may be a ploy to keep investors on board.