Anthropic’s rapid 2026 Claude 5-series releases, including Opus 5.5 and Fable 5.1, have driven the leading HLE scores to 65% as of October, establishing Anthropic’s edge on the expert-level benchmark of 2,500 closed-ended questions across advanced domains. Traders see continued momentum from Anthropic’s focus on reasoning depth and tool use, outpacing GPT-6 variants and Gemini models that trail at 57-59%. With the year-end window still open for additional model iterations, fine-tunes, or agentic enhancements before the January 2027 resolution, the market prices a strong likelihood of 55%+ while assigning lower odds to 65%+ or higher, reflecting both proven progress and the benchmark’s resistance to quick saturation.