Tyler Johnston
I was avoiding weighing in on this, but does anyone at OpenAI dispute that: (1) They heard rumors of a Millennium Prize breakthrough (the rumors I heard were that it was about Naiver-Stokes specifically, from Anthropic, and under external review. Who knows if they knew this, the blog posts suggests they didn't, but even so...) (2) They then spent millions of dollars in a week-long sprint to... scoop the (unknown) authors who made the breakthrough? Or at least push hard in that direction knowing there was a chance they would do so? (3) They then learned that Levant/Tristan were the authors in question and intended to offer them co-authorship? For a result that was presumably already achieved? Then they learned Levant/Tristan only had a resolution of the forced Euler problem (i.e. a key step on the way to a Navier-Stokes solution, but not the full milestone) so they published alone? If all the above is true, then sure. OpenAI independently found a solution to the Naiver-Stokes equation. But they seemingly did it for maximally icky reasons. I don't know the professional norms in math but something about this reeks of... silicon valley mindset/zero-sum thinking. Why not spend the millions of dollars pursuing a totally new problem, rather than directly competing with progress on a problem that you know is about to be announced, even if you don't know the exact shape or origin of the announcement? Even if you want to explore problems in the region of "millennium prize," why not give it two weeks to see what the pending result is so you can build off of it, not scoop someone, not waste millions on potentially duplicative efforts, etc. Seems like a waste, unless the goal is to beat someone else to the announcement, i.e. "I don't want to live in a world where someone else solves millennium prize problems better than we do."
Noam Brown
Yes, this result cost millions of dollars. But remember that when @OpenAI announced o3 it cost ~$500,000 to score 87.5% on ARC-AGI 1. Today, Astra scores higher for ~$20. In 2025 it took us and GDM an enormous amount of compute to achieve IMO gold. For the 2026 IMO, anyone with a $20/month ChatGPT subscription could do it. Massively scaling test-time compute gives us a glimpse of the future. I believe that a year from now everyone will have an AI at their fingertips capable of solving problems of this caliber.