Google and Anthropic models dominate recent Text Arena Math rankings, with Gemini 4 Argon variants posting the highest scores around 1528 and Claude Opus/Fable series clustered just behind at 1524–1514, creating the tight 30% and 29.5% market-implied odds for second place by year-end. These outcomes reflect sustained high-effort reasoning improvements and consistent leaderboard positioning over the past month rather than isolated benchmark spikes. Meta trails at 13% amid lower model ranks, while Chinese labs and OpenAI sit further back with scores in the 1490s or below. Fresh releases or high-compute updates from either leader before December could still flip relative standings in this narrow contest.