Meta Launches Muse Wonder Models With Active Memory And Focus Capabilities

Meta has launched two new artificial intelligence models, Muse Wonder and Muse Wonder Full, which the company says represent a major leap toward personal superintelligence. According to Engadget, these models introduce active memory and active focus abilities, allowing the systems to retain relevant details across conversations and persist through difficult tasks. This release follows more than a year of rebuilding Meta’s AI efforts, with the company stating it has now caught up to the industry frontier.

What new capabilities do these models offer?

The new models add active memory, which Meta describes as the ability to carry what matters to users across conversations and bring back details relevant to the task at hand. Active focus allows the agent to keep working toward a goal when a task gets hard. According to Meta, these features help people accomplish more over long periods of time, particularly in navigating legal, financial, and health-related tasks. The company also claims the models deliver frontier coding performance.

How does this relate to Meta’s broader AI strategy?

Meta expects to spend as much as $145 billion this year to fuel its ambitious AI plans, a significant investment in compute and scientific work. In an interview with the Financial Times, CEO Mark Zuckerberg stated that getting to this point required a large amount of compute and investment, along with rigorous scientific work from the company’s superintelligence team. The new models are a notable milestone in this broader effort to rebuild and advance Meta’s AI capabilities after a period of strategic restructuring.

Who can access these models and what do they cost?

Muse Wonder is available via Meta’s API and in Muse Code, priced at $4 per million input tokens and $20 per million output tokens. Muse Wonder Full is available to people who subscribe to the $50 per month power usage tier or the $200 per month maximum usage tier of Muse Code. Meta plans to bring Wonder to its personal agent Muse, beginning with paid subscribers. The rollout to subscribers is happening now, providing early access to these advanced features for paying customers.

What does this mean for business operations?

For organizations evaluating AI tools, the introduction of active memory and focus suggests a shift toward more persistent and context-aware agents. This could impact how businesses integrate AI into workflows that require long-term context retention, such as legal document review or complex financial analysis. The pricing structure indicates a tiered approach, with higher costs for more intensive usage, which businesses should factor into their operational budgets. The availability via API allows developers to integrate these capabilities into existing systems, potentially enhancing productivity in specialized domains.

What to do and how to stay safe: Meta

• Review current AI vendor contracts to compare pricing structures for input and output tokens against Meta’s new rates.
• Evaluate whether active memory features align with your organization’s need for long-context retention in legal or financial workflows.
• Monitor the rollout of these models to the personal agent Muse to assess integration opportunities for paid subscriber tiers.
• Assess the security implications of AI agents retaining personal details across conversations and ensure data privacy policies are updated accordingly.

Step-by-step guide: How AI Risk Management Works: The Hidden Failure Points

General security guidance from the Sandbox Dispatch newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

What are the pricing details for Meta’s new Muse Wonder models?

Muse Wonder costs $4 per million input tokens and $20 per million output tokens via API and Muse Code. Muse Wonder Full requires a $50 or $200 monthly subscription to Muse Code.

What is active memory in the context of these AI models?

Active memory is the ability to carry what matters to users across conversations and bring back details relevant to the task at hand, helping people accomplish more over long periods.

How much is Meta planning to spend on AI this year?

Meta expects to spend as much as $145 billion this year to fuel its ambitious AI plans, including compute and investment in its superintelligence team.