Meta leads with a 62% implied probability for third place on the Text Arena math leaderboard by end of October, driven by its October 2 release of six research papers where Muse Spark models assisted in solving open problems in group theory, differential equations, and probability. Traders view this as strengthening Meta’s competitive positioning relative to Google and Anthropic, which currently rank first and second on the arena.ai math board, while OpenAI sits just ahead of Meta in fourth. Lower odds for Z.ai, Moonshot, and others reflect their solid but less prominent recent math-specific gains. With resolution imminent, any late-October model updates or benchmark shifts could still alter the final lab ordering.