Meta on Monday released its first models from Superintelligence Labs, Muse Wonder and Muse Wonder Full. According to data released by Meta, Muse Wonder outperformed OpenAI's GPT-6 Astra and Anthropic's Claude Opus 5.5 in memory, legal, and marathon programming tests. Specifically, Muse Wonder Full scored 59.1 on Meta's internal Personal Superintelligence Index, higher than Claude Opus 5.5's 55.2 and GPT-6 Astra's 54.5. In the SWE-Marathon programming test, Wonder Full scored 57.5, while Opus 5.5 scored 46.3 and Astra scored 42.5. In Harvey's legal representation benchmark test, Wonder Full scored 13.1, Astra scored 8.6, and Opus 5.5 scored 4.2. Additionally, in Meta's MemoryBench test, the Wonder model scored approximately 85.5, Opus 5.5 scored 81.5, and Astra scored 81.4. However, independent ranking agency Artificial Analysis's overall assessment still places Anthropic's Opus 5.5 in the top spot, with Muse Wonder Full tied for second with the more affordable Claude Sonnet 5.5.