Google's Gemini 4 Carbon overtook OpenAI's GPT-6.1 Sol as the MathArena leader in mid-November 2026, posting an expected performance of 85.4 percent on the researcher-run benchmark of competition-level math problems. Anthropic's Claude Fable 5.2 sits in second at 82.8 percent, while several other frontier large language models from OpenAI, Google, and Anthropic cluster in the low-to-mid 80s. The benchmark emphasizes multi-step reasoning accuracy alongside inference cost, with recent model releases and reasoning configurations driving the rapid shifts in rankings. With six weeks remaining until year-end, any new releases or capability updates from the major labs could alter the leaderboard before December 31.
This market will resolve to "Yes" if any model on the Arena.AI Leaderboard (arena.ai/leaderboard/text) reaches at least the specified Arena Score on the "Leaderboard" tab for "Math" by December 31, 2026, 11:59 PM ET. Otherwise, this market will resolve to "No". Results from the "Score" column under the "Text Arena | Math" Leaderboard tab at https://arena.ai/leaderboard/text/math-no-style-control with style control off will be used to resolve this market. The resolution source for this market is the Chatbot Arena LLM Leaderboard found at arena.ai/leaderboard/text. If this resolution source is unavailable at check time, this market will remain open until the leaderboard comes back online and will resolve based on the first check after it becomes available. If permanently unavailable, this market will resolve to "No".