OpenAI’s October 2026 release of 722 AI-generated papers from an unreleased internal frontier model, claiming progress on roughly 377 open problems across number theory, algebra, topology, and geometry, has sharply lifted trader expectations for additional breakthroughs before year-end. Many proofs include Lean formalizations for verification, though mathematicians continue independent review amid concerns over exposition and potential errors. This follows the company’s September claim of resolving a Navier-Stokes variant, one of the Millennium Prize Problems. Competitive efforts from Google DeepMind and Anthropic on Erdős problems and formalizations add momentum, while the market-implied odds favor outcomes like the Hadamard Matrix Conjecture. Key catalysts ahead include further Lean checks, lab announcements, and any public model releases that could accelerate resolution of remaining conjectures.