Google and Anthropic models currently anchor the top of Text Arena math leaderboards, with Gemini 4 Argon variants posting the highest Elo scores near 1530 while multiple Claude Opus and Fable releases sit within 10–15 points, creating a narrow gap that keeps their second-place probabilities closely matched at 33.5% and 26.5%. Recent flagship updates from both labs, including enhanced reasoning modes and longer context handling, have sustained this positioning through October 2026. Meta’s Muse Spark series, alongside open-weight contenders from Z.ai, Alibaba’s Qwen, and Xiaomi, trail but could close ground with targeted math fine-tuning or new releases before year-end. Trader consensus reflects these incremental capability gains and the short runway to December resolution.