Meta's Muse series models, including Muse Spark 1.1 and Muse Wonder, have posted competitive but non-leading results on Humanity's Last Exam variants in late 2026, with scores ranging from roughly 49% to 62% across tool-free and tool-enabled leaderboards. Anthropic's Claude Opus 5.5 and Fable 5.1 variants currently top most public rankings at 59-64%, followed closely by OpenAI's GPT-6 Astra and Google's Gemini iterations, reflecting faster capability gains from closed labs. Key drivers include Meta's focus on efficient, widely available large language models versus rivals' emphasis on frontier reasoning benchmarks. Upcoming model updates or partnerships could shift relative standings before year-end resolution, though historical patterns show rapid leaderboard turnover with new releases.