Meta's pitch for Muse Wonder and Muse Wonder Full, the two models it released Monday, comes down to two features it calls Active Memory and Active Focus. The first carries personal context from one conversation to the next and pulls up the relevant details when a task needs them. The second is meant to keep the model working on hard goals: trying other approaches when something fails, asking for clarification, and checking in before important actions.
"We 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 research scientist Kavya Kopparapu said in a launch video, as quoted by TechCrunch. Meta says Muse Wonder beats Fable, Opus and Astra on MemoryBench, a memory benchmark it highlighted, though that comparison comes from its own testing.
There's an outside number too. Artificial Analysis scored Muse Wonder at 52 and Wonder Full at 56 on its Intelligence Index v4.3 at max reasoning effort. Muse Spark 1.3 scored 48, and it shipped just six weeks earlier.
Both models were trained at Meta's Prometheus data center in Ohio. TechCrunch calls this Meta's first frontier model release since Alexandr Wang restructured the company's AI work in August 2025. Meta also says Wonder Full does better on unfamiliar tasks than its Muse Spark 1.3 trend line predicts as test-time compute goes up, and it credits a bigger base model.
Access is split. Muse Wonder is live in the Meta Model API at $4 per million input tokens and $20 per million output tokens, and in Muse Code. Wonder Full is only in Muse Code, on the Power Usage and new Maximum Usage subscription plans, with no API access. Meta says Wonder will also come to the Muse app, subscribers first.
A model that's built to remember your messages, calendar and documents needs to be careful with them, and Meta knows it. The company calls Muse Wonder its most aligned model yet, says it has its strongest safeguards against chemical, biological, radiological, nuclear and cyber misuse, and says a fuller safety report is coming soon.
— via TestingCatalog.