The models add “active memory” and “active focus” abilities.
After spending more than a year and tens of billions of dollars rebuilding its AI efforts, Meta says that it has caught up with the frontier. The company introduced its two latest AI models, Muse Wonder and Muse Wonder Full, both of which represent “a major leap toward personal superintelligence.”
According to Meta, the models are particularly adept at navigating legal, financial and health-related tasks, and are able to deliver “frontier coding performance.” Both models also add “active memory” and “active focus,” which Meta says will help people accomplish more over long periods of time.
That’s notable because Meta says it plans to bring Wonder to its personal agent Muse, beginning with paid subscribers. Active memory, according to the company, is “the ability to carry what matters to you across conversations and bring back the details that are relevant to the task at hand” while active focus will allow an agent to “keep working toward your goal when a task gets hard.”
The models are a notable milestone for Meta, which has said it expects to spend as much as $145 billion this year to fuel its ambitious AI plans. In an interview with the Financial Times, CEO Mark Zuckerberg said that “getting to this point both required a large amount of compute and investment, and a lot of very rigorous scientific work” from the company’s superintelligence team.
The new models are rolling out now to subscribers. Muse Wonder is available via Meta’s API and in Muse Code. It’s 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/month “power usage” or $200/month “maximum usage” tiers of Muse Code.