OpenAI’s early October 2026 release of 722 AI-generated papers advancing or resolving hundreds of open problems, including partial progress on multiple Millennium Prize challenges such as Navier-Stokes and quasi-Riemann variants, has sharply accelerated trader focus on which conjectures could receive qualifying full-solution announcements by year-end. Labs including DeepMind with formal Lean-based agents and Anthropic have similarly demonstrated autonomous resolution of longstanding Erdős problems and other research-level questions, underscoring rapid gains in large language model reasoning combined with verification tools. Key swing factors include independent mathematician verification of formal proofs, the pace of unreleased frontier model iterations, and whether labs issue explicit “solved” statements meeting market criteria before December 31.