Meta’s Muse Spark series currently posts the strongest HLE results among its releases, with Muse Spark 1.1 reaching approximately 62% on select leaderboards while Muse Spark 1.3 sits near 48-49% on Artificial Analysis and similar evaluations. These scores place Meta competitive with mid-tier frontier systems but behind Anthropic’s Claude Opus 5.5 and Fable variants (59-65%) and select OpenAI GPT-6 configurations. Rapid benchmark gains across labs—driven by larger models, improved reasoning chains, and tool use—set the pace, yet Meta’s open-weight strategy and efficiency focus may constrain peak scores relative to closed competitors. Traders are watching Meta’s remaining 2026 release cadence, including potential Llama or Muse Spark updates before year-end, for any new high-water mark on the 2,500-question HLE benchmark.