Meta Launches Muse Wonder, Its First Frontier AI Model

Summary

Meta’s Superintelligence Lab released Muse Wonder and Muse Wonder Full on October 12, 2026, its first frontier models and larger, more capable successors to lightweight, low latency Muse Spark. Meta’s model card says both deliver stronger benchmarks and complex task performance, including coding, comparable with Fable 5.1 and GPT-6 Astra. Wonder also leads Fable, Opus and Astra on MemoryBench, which tests continual learning and memory construction. Meta calls this ability to retain and synthesize user information “active memory.” Research scientist Kavya Kopparapu said recall is designed to persist across days, weeks, months and beyond.

Wonder costs $4 per million input tokens and $20 per million output tokens through the Meta Model API, roughly matching GPT-6 Sol or Opus 5.5. Wonder Full is restricted to Muse Code subscribers and will not be offered through the API, while both models will power the Muse agent. Trained at Meta’s Prometheus data center in central Ohio, which came online earlier in 2026, they are Meta’s first frontier release since Scale AI founder Alexandr Wang reorganized its AI operation in August 2025. Meta describes the launch as its next scaling step toward personal superintelligence.