OpenAI’s October 6 disclosure that an internal frontier model produced verified solutions to more than 100 long-standing open math problems has sharply lifted trader expectations for the FrontierMath: Open Problems market. Epoch AI’s independent Open Problems set, which only counts solutions once officially classified, already lists several AI or human-AI resolutions since the July launch, and the rapid saturation of Tier 4 (GPT-6 Astra at 97.6 %) plus new Erdős and HorizonMath benchmarks signals sustained momentum. Competitive pressure from Anthropic, Google, and others, combined with improving reasoning scaffolds and Lean verification, makes additional Epoch-classified solves before year-end the base case, though exact timing hinges on when Epoch next updates its public ledger.