Anthropic’s November 9 release of Claude Fable 5.2 has pushed the company to the top of most independent benchmarks, including Artificial Analysis Intelligence Index and BenchLeader scores, ahead of Google’s recently generally available Gemini 4 Carbon. Traders on Polymarket nevertheless assign Google roughly 70% implied probability of owning the year-end #1 model on the Chatbot Arena leaderboard, reflecting expectations of further DeepMind progress or Arena-specific performance before December 31. OpenAI’s GPT-6 series trails on current evaluations, while Meta and xAI remain further back. With only seven weeks remaining, any additional frontier releases or benchmark shifts could alter the tight competitive positioning among the leading labs.