Meta has officially entered the frontier model race with the release of Muse Wonder and Muse Wonder Full. Developed by Meta Superintelligence Labs and trained at the company gigawatt-scale Prometheus data center, these systems represent Meta's most capable artificial intelligence models to date. Unlike models designed primarily for quick chat responses or one-off queries, the Muse Wonder series is engineered specifically to handle complex, long-running autonomous agent tasks.
The launch marks a significant milestone in Meta's fourteen-month overhaul of its artificial intelligence infrastructure, building upon the predictable scaling framework previously established by Muse Spark. Available immediately through the Meta Model API, AI/ML API, and Muse Code, these models are designed to bridge the gap between casual assistance and heavy-duty personal and professional execution.
At the core of the Muse Wonder series are several architectural breakthroughs tailored for autonomy. First is Active Memory, a system that allows the models to retain and synthesize user-provided information across conversations, actively retrieving relevant details when needed. This is paired with Active Focus, a mechanism that enables the models to pursue challenging goals over extended periods, try alternative approaches, work through blockers, and adapt to new information without losing the thread of the objective.
The models also boast state-of-the-art domain expertise across specialized fields such as law, finance, and health, performing strongly on benchmarks like Harvey Legal Agent Benchmark (LAB). Furthermore, the models deliver frontier coding performance, supporting multi-file codebases and complex software engineering tasks tested on evaluations like SWE-Marathon.
For founders, builders, and business leaders, the arrival of Muse Wonder signals a shift away from conversational AI and toward true digital agency. As AI systems become capable of maintaining long-term context and executing multi-step workflows autonomously, the opportunity shifts from prompting chatbots to orchestrating digital workers that can handle software engineering, legal analysis, and complex operational processes independently.