OpenAI’s recent solution to the Navier-Stokes Millennium Prize problem in September 2026, achieved via thousands of internal AI agents on a frontier model in roughly 88 hours, has anchored trader sentiment at 80.5% implied probability. The company followed with a October release of 722 papers containing hundreds of new proofs across major open questions, including progress toward variants of the Riemann hypothesis, underscoring its scaled reinforcement learning approach and compute advantage. Anthropic has delivered strong supporting results, such as formalizing Fermat’s Last Theorem and advancing percolation theory conjectures with Claude, yet lacks a comparable Millennium Prize resolution or equivalent volume of verified breakthroughs. Key near-term catalysts include independent Lean verifications of OpenAI’s claims, any joint or rival announcements, and next-model training runs that could accelerate either lab’s timeline on remaining problems like the Riemann hypothesis.