Meta Platforms has introduced its latest artificial intelligence models, Muse Wonder and Muse Wonder Full. The company stated these models represent its most advanced capabilities to date for agentic tasks, covering personal and professional applications. Muse Wonder is now accessible to developers through the Meta Model API and Muse Code. Muse Wonder Full is available to users subscribed to Muse Code.
These new models are engineered with features named Active Memory and Active Focus. Active Memory allows the AI to retain information across multiple conversations. Active Focus enables the models to persist with challenging tasks, explore different solutions, and request clarification when necessary. Meta trained the Muse Wonder models with a focus on safety, including protocols to prevent sharing private information without consent and to decline requests for harmful activities. The company developed these models under its Meta Superintelligence Scaling Framework.
Meta CEO Mark Zuckerberg described Muse Wonder as a "family of frontier models with state-of-the-art performance for personal and professional agentic work." The models demonstrate expertise in domains such as law, finance, and health, and offer advanced coding capabilities. Meta claims that internal benchmarks show Muse Wonder Full outperforms leading models from OpenAI and Anthropic in areas like memory retention, financial analysis, and long-horizon coding tasks. However, the company has not released comparative benchmark results with rival models in its announcement.
The Muse Wonder models are intended to power Meta's existing Muse AI agent, with initial rollout for subscribers. This integration aims to enhance the agent's performance in real-world tasks, such as navigating insurance claims, interpreting legal leases, and building applications. Meta indicated plans to eventually develop specialized models for broader use within its Muse products. The company has undergone a significant overhaul of its AI efforts over the past 14 months, establishing Meta Superintelligence Labs to drive its AI development. This strategic shift follows investments in AI infrastructure and talent, including the acquisition of a stake in Scale AI and the leadership of Alexandr Wang at Meta Superintelligence Labs.
Meta has offered Muse Code subscriptions with different tiers, including "Everyday Usage," "High Usage," and "Power Usage," providing varying levels of access and prompt allowances. The company positions these releases as a significant step toward achieving personal superintelligence.