Anthropic’s April 2026 releases of Claude Opus 4.7 and the Mythos Preview have propelled its models to leading scores of 35–65% on Humanity’s Last Exam, a 2,500-question benchmark testing frontier-level reasoning across mathematics, science, and humanities. These results, often achieved with tool use and extended chain-of-thought, surpass prior Claude versions and edge out competitors such as OpenAI’s GPT-5 series and Google’s Gemini 3.1 Pro. Trader focus now centers on whether a publicly listed Claude variant can sustain or exceed these thresholds through June 30 amid ongoing evaluation updates and rapid model iterations. Potential catalysts include any Claude 5 preview announcements or benchmark refreshes at upcoming AI conferences, while discrepancies in prompting methods or leaderboard methodology remain key variables.