Why this score Meta's AI division; principal open-weight publisher among US labs. Authoritative primary for its own releases; motivated party, with past benchmark presentations contested by independent evaluators.
The Muse Wonder series achieves state-of-the-art results across key benchmarks for personal and professional agents, memory, and coding.
Partly holds. Artificial Analysis puts Muse Wonder Full in its top three on six of ten index tests and level first on its coding-agents index (68, with Claude Sonnet 5.5), but second overall at 56 against Claude Opus 5.5's 58, and calls knowledge a relative weakness.
Meta will 'soon' release open weights for its flagship Muse Spark 1.2, extending its open-weight pivot from the small Muse Glimmer to a frontier-band model.
Half kept. On 15 October Meta published weights for a frontier-band Muse Spark model, Muse Spark Open (816B parameters; Artificial Analysis score 39), but under Meta's own restrictive licence, and no source identifies it as Muse Spark 1.2.
We believe an open approach is the right one for the development of today's AI models, especially those in the generative space where the technology is rapidly advancing. By making AI models available openly, they can benefit everyone.
A development-philosophy statement, not falsifiable — but measurable against Meta's conduct, which has now reversed twice. Its Muse flagship line shipped proprietary (Muse Spark 1.1 and 1.2), then in August 2026 Meta returned to open weights, releasing Muse Glimmer and pledging to open Muse Spark 1.2. The 2023 argument stands as the founding statement of the strategy Meta abandoned and has now resumed.