Recent breakthroughs on Epoch AI’s FrontierMath: Open Problems benchmark have accelerated trader sentiment toward near-term resolutions. GPT-6 Astra and OpenAI’s internal models have delivered verified solutions to multiple open research problems in September 2026, including a “major progress” level conjecture unsolved since 2017 and key aspects of the Navier–Stokes equations. These results follow earlier 2026 milestones such as the 2-adic Galois group and hypergraph problems, with independent Lean formalizations and mathematician validation confirming the advances. OpenAI’s rapid iteration—training a new model in late August that cleared over 100 high-level problems—has outpaced competitors like Anthropic’s Claude Fable series on Tier 4 challenges. With Epoch AI actively updating solved statuses and frontier labs racing to demonstrate further capability gains before year-end, market-implied odds reflect strong expectations of at least one additional official classification by late September or December.