Meta’s Muse Spark series, including the 1.1 variant scoring 52–62% across October 2026 leaderboards depending on tool use and verification, currently trails Anthropic’s Claude Opus 5.5 and Fable 5.1 at the frontier of Humanity’s Last Exam (HLE), a 2,500-question expert benchmark testing advanced reasoning beyond saturated tests like MMLU. Rapid capability gains across labs—driven by scaling, post-training, and tool integration—have lifted top scores from the mid-40s earlier in the year, creating a narrow window for Meta to close the gap before year-end model drops. Traders weigh Meta’s release cadence and open-weight strategy against competitive pressure from closed frontier systems, with any new Muse iteration or benchmark update serving as the key near-term catalyst through December.