Anthropic’s rapid iteration on its Opus line has positioned the company’s frontier models at the top of Humanity’s Last Exam leaderboards. The September 22, 2026 release of Claude Opus 5.5 delivered immediate leadership on tool-augmented HLE (67.7%) and Artificial Analysis HLE (61.4%), extending the pattern set by Opus 5 in July and internal variants such as Fable 5.1. These results reflect sustained gains in agentic coding, long-horizon reasoning, and knowledge work that directly map to the benchmark’s expert-level questions. Traders are watching for any announcement of a follow-on Opus release or public benchmark update, as Anthropic’s cadence and consistent outperformance on closed and tool-enabled evaluations continue to shape market-implied odds around debut timing and score thresholds.