Recent Gemini 4 Argon releases have propelled Google to a narrow lead on Text Arena's Math leaderboard through strong crowd-sourced performance across complex problems, yet Anthropic's Claude Opus and Fable variants remain highly competitive in mathematical reasoning benchmarks and human preference votes. The near-even market-implied odds around 48-51% reflect this tight contest, with traders viewing outcomes as sensitive to any late-October model updates or fine-tunes that could shift arena rankings. Open-weight contenders like those from Moonshot or Z.ai show promise on select evals but trail frontier closed models in overall arena strength.