Google's Gemini models, particularly the 3.7-flash-high and 3.6-flash-high variants, have secured multiple top-five positions on the Text Arena Math leaderboard as of late August 2026, establishing the lab as the clear market favorite for second place behind Anthropic. This positioning stems from sustained gains in pure mathematical reasoning benchmarks and competitive Elo scores against frontier models from OpenAI and Alibaba, whose recent math-focused releases and agentic tools have not yet translated into equivalent arena dominance. Traders assign Google an 84% implied probability because its consistent mid-tier model performance creates a wide gap over lower-ranked labs, though tight clustering among Chinese and U.S. contenders leaves room for last-minute shifts before the September 30 resolution.