Recent OpenAI releases have driven trader sentiment on AI math progress in 2026, including a September claim resolving aspects of the Navier-Stokes equations and an early October dump of 722 manuscripts from an internal frontier model addressing 377 open problems across algebra, number theory, topology, and computer science. These include advances on the quasi-Riemann hypothesis, Unique Games Conjecture, and Hodge conjecture for certain abelian varieties, with many proofs formalized in Lean though independent verification remains ongoing. Smaller labs like doubleAI have also delivered low-cost solutions to longstanding graph theory problems, highlighting competitive dynamics between scale-focused frontier labs and efficient agent systems. With the market resolution tied to official announcements by year-end, key catalysts include further model releases, rigorous peer review of the October claims, and any Millennium Prize-level confirmations before December 31.