Anthropic holds the strongest market-implied odds at 60.5% for the best LiveBench coding model by end of December, driven by its Claude Fable 5.2 and Opus 5.5 variants posting leading scores on recent LiveBench coding and research leaderboards, often 2–5 points ahead of Google Gemini 4 Carbon and OpenAI GPT-6 models. These releases delivered measurable gains in agentic coding tasks and contamination-resistant benchmarks, reinforcing trader consensus around Anthropic's edge in practical software engineering performance. Google (17.7%) and OpenAI (13.0%) trail due to narrower gaps on the same evaluations, while lower-probability entrants like Xiaomi or DeepSeek lack comparable frontier results. With roughly seven weeks remaining, new model drops or benchmark refreshes could still alter positioning before resolution.
基于Polymarket数据的AI实验性摘要。这不是交易建议,也不影响该市场的结算方式。 · 更新于1H6H1D7D30DALL
View resolved






















警惕外部链接哦。
警惕外部链接哦。
常见问题