Meta Introduces Muse Wonder Models to Advance Agentic AI

Meta has introduced Muse Wonder and Muse Wonder Full, two advanced artificial intelligence models designed to perform complex personal and professional tasks, as the company seeks to strengthen its position in the increasingly competitive AI market.

Meta introduces Muse Wonder and Muse Wonder Full, new AI models designed to handle complex tasks and power future experiences in its Muse AI assistant. [Image: Mark Zuckerberg Facebook]

The launch addresses two major questions surrounding Meta's AI strategy: whether its revamped Superintelligence Labs division can deliver more competitive models and whether its substantial investment in computing infrastructure will generate sufficient returns.

Meta spent much of 2026 rebuilding its AI capabilities.

The company introduced the Muse Spark family of large language models in April, followed by its personal AI agent, Muse, last month.

Recently, Meta expanded the lineup with Muse Wonder and Muse Wonder Full, which it describes as its most capable models to date for agent-based tasks.

Although Muse Spark also supports agentic capabilities, Meta says the new models are larger and more capable, allowing them to tackle more demanding tasks, albeit with potentially longer processing times.

Muse Wonder was trained at Meta's Prometheus data center and is designed to support tasks such as navigating insurance claims, interpreting leases, handling financial information and building applications.

The models are also expected to power future features in the Muse assistant and support the development of specialized AI systems.

In its announcement, Meta said the models performed competitively against leading systems from OpenAI and Anthropic on selected benchmarks.

The company highlighted two capabilities: Active Memory, which allows models to retain relevant context across conversations, and Active Focus, which helps them persist when initial approaches fail.

The launch is part of Meta's broader effort to turn its AI investment into new sources of revenue.

The company has historically relied heavily on digital advertising, and it is seeking to expand its consumer AI ecosystem through products and subscriptions.

However, the commercial payoff remains uncertain. Meta will need to attract paying users and demonstrate that its AI products can generate meaningful revenue relative to the substantial costs of developing and operating advanced models.