Anthropic’s latest Claude variants, including Fable 5, Mythos 5, and Opus 4.8, currently lead Humanity’s Last Exam leaderboards with scores ranging from 53% to 64.5% under varied testing conditions such as adaptive reasoning and tool use. These results reflect iterative gains in large language model reasoning on the 2,500-question benchmark spanning expert-level math, science, and humanities topics. Competitive pressure from OpenAI’s GPT-5 series and Google’s Gemini 3.1 Pro, which trail but remain close, continues to accelerate development cycles across labs. With only two weeks remaining until the June 30 resolution date, trader sentiment hinges on whether Anthropic ships a meaningful update or further optimization before the cutoff, given typical release timelines and the benchmark’s resistance to rapid saturation.