Anthropic’s September 2026 launch of the Claude 5.5 family, including Opus 5.5, has driven current trader sentiment by posting leading scores of 61–67.7% on multiple Humanity’s Last Exam variants, including tool-augmented and Artificial Analysis harnesses. These releases emphasize advanced reasoning, agentic workflows, and expert-level performance across the 2,500-question benchmark spanning math, sciences, and humanities. Claude models currently outpace GPT-6 and Gemini counterparts on most public leaderboards, though exact results vary by protocol, tool access, and subsets like HLE-Diamond. With resolution still months away, additional model updates, benchmark refinements, or competitive jumps from rival labs remain key swing factors that could lift the highest Claude score toward or beyond 65–70%.