OpenAI's October 2026 release of 722 AI-generated math papers, produced by an unreleased internal frontier model, has become the dominant driver of trader sentiment on which problems large language models will fully solve by year-end. The batch includes claimed advances on several Millennium Prize Problems, such as partial progress toward the Riemann hypothesis via a quasi-variant and work tied to Navier-Stokes, alongside results in number theory, algebraic geometry, and graph theory, many formalized in Lean for verification. Other labs, including doubleAI and prior DeepMind AlphaProof efforts on IMO-level problems, underscore competitive dynamics in automated theorem proving. Verification by mathematicians remains ongoing, with some retractions already noted, while the short timeline to December 31 creates uncertainty around formal announcements required for market resolution.