OpenAI announced that its AI system solved the Navier-Stokes problem, one of the seven Millennium Prize Problems. The company utilized 10,000 AI agents working for 88 hours to reach the proof. The solution was generated by an unreleased model, which OpenAI claims is significantly more capable than its GPT-6 Astra system.
Following the generation of the proof, the GPT-6 Astra model spent 17 hours verifying the logic. While the Clay Mathematics Institute offers $1 million for solving any of the seven Millennium Prize Problems, OpenAI stated it will not claim the prize money.
The announcement sparked a dispute regarding the origin of the work. NYU professor Tristan Buckmaster and Anthropic mathematician Levent Alpöge noted that their research was previously included in OpenAI’s Codex model. Buckmaster suggested this might have exposed their unpublished findings to OpenAI’s team.
OpenAI researcher Sebastien Bubeck denied that the company accessed or utilized the unpublished work of the mathematicians. However, OpenAI acknowledged it cannot rule out the possibility that the researchers' product usage influenced the training data of its model.
The controversy emerges as OpenAI prepares for a reported $1 trillion public listing. Meanwhile, competitor Anthropic is also pursuing a significant valuation through its own planned IPO.
