OpenAI’s October 6 release of 722 manuscripts from an unreleased internal frontier model, covering 372 result families with hundreds of claimed solutions to open mathematics problems and many Lean-verified proofs, has sharply accelerated trader expectations for additional FrontierMath: Open Problems entries to be marked solved by Epoch AI. This follows September’s reported resolution of a Millennium Prize problem (Navier-Stokes) and earlier GPT-6 Astra successes on Erdős problems, underscoring OpenAI’s lead in autonomous large-language-model reasoning over rivals such as Anthropic and Google models. Mathematicians continue parsing the new results for novelty and correctness, while Epoch AI’s benchmark updates and any further formal verifications represent the key near-term catalysts that could confirm the next solved problem and drive market resolution.