Meta Superintelligence Labs released Muse Wonder and Muse Wonder Full, their latest “most-capable” AI models bringing together Active Memory and Active Focus with specialized domain knowledge.
Announced by CEO Mark Zuckerberg, the pair represents Meta’s first true frontier-class models and a major step toward “personal superintelligence” after roughly 14 months of intensive rebuilding of its AI stack.
Introducing Muse Wonder, our new family of frontier models with state of the art performance for personal and professional agentic work. https://t.co/NKHrybrGKQ
— Mark Zuckerberg (@finkd) October 12, 2026
According to Meta, Muse Wonder and Muse Wonder Full emphasize agentic performance on personal and professional fronts with core innovations Active Memory that allows model to retain and retrieve relevant user context and Active Focus that sustains multi-step effort by adapting, clarifying and checking progress.
“Trained at our gigawatt-scale Prometheus data center, Muse Wonder and Muse Wonder Full bring together the work of teams across Meta to deliver frontier capabilities for complex tasks at work and in everyday life,” read a statement from Meta.
Meta has claimed Muse Wonder and Muse Wonder Full with state-of-the-art results on agentic work, memory and coding benchmarks with competitive or leading scores against AI models such as Claude Fable 5.1, Claude Opus 5.5 and GPT-6 Astra on indices covering legal, finance, health and long-horizon software work.
Meta has announced that Muse Wonder has been made available immediately through Meta Model API and Muse Code while the larger Muse Wonder Full has been restricted to Muse Code subscribers.
Also Read: Meta Launches Muse Glimmer 30B Open-Weight Local AI Model
