Meta's Superintelligence Lab released its first frontier AI models, Muse Wonder and Muse Wonder Full, on Oct. 12, built for complex agentic work like insurance claims, leases and coding, the company said.
The models center on what Meta calls active memory, the ability to retain and synthesize information a user provides over days, weeks or months, and active focus, the ability to persist through a difficult task and recover from its own errors, according to Meta's announcement.
Wonder costs $4 per million input tokens and $20 per million output tokens through the Meta Model API, TechCrunch reported, pricing it close to rival frontier models from other labs. Wonder Full is available only to Muse Code subscribers and is not offered through the API.
The models were trained at Meta's Prometheus data center in Ohio, which opened earlier this year, TechCrunch reported. Meta research scientist Kavya Kopparapu said the team "put a tremendous amount of effort into making sure your model can remember what's important to you over days, weeks, months, and beyond."
Meta had trailed OpenAI, Anthropic and Google on frontier benchmarks through most of the generative AI boom. A model priced to compete and built around memory that persists across sessions gives builders another option for agents that need to carry context across long-running tasks.