Anthropic’s Claude Fable 5.1, released September 1, 2026, currently leads the Humanity’s Last Exam (HLE) leaderboard at 65% in no-tools evaluations, ahead of Opus 5 and Mythos 5 variants, reflecting strong gains in multidisciplinary expert reasoning across the 2,500-question benchmark. Rumors point to a Fable 5.2 or 5.5 pretraining run arriving in late September or early October, though Anthropic has not confirmed any timeline or HLE-specific debut strategy. Recent HLE-Diamond and with-tools leaderboards show Opus 5.5 pulling ahead at 67.7%, underscoring competitive pressure from OpenAI’s GPT-6 Astra and Google’s Gemini 3.8 series. Traders are watching for official announcements or API identifiers that could signal whether the next Fable iteration targets HLE as its flagship demonstration of frontier capabilities.