Meta rolled out two models on October 12: Muse Wonder and the higher-tier Muse Wonder Full. The split is less about naming than access. On the Meta announcement, Wonder is presented as the version developers can use through the Meta Model API and in Muse Code, while Wonder Full is reserved for paid Muse Code tiers.
For the API-available model, Meta lists a 1M-token context window and usage-based pricing. Wonder Full is instead tied to Muse Code’s Power Usage and Maximum Usage subscriptions, which Meta prices at $50 per month and $200 per month.
That makes this both a model launch and a packaging decision. Teams building agents or coding workflows can wire Muse Wonder into existing API-based systems, but getting access to Wonder Full may mean shifting some work into Meta’s own subscription environment rather than treating it like a standard model endpoint.
Meta also says Wonder Full performs ahead of several coding and agent benchmarks in third-party evaluations. Those performance claims come from Meta’s launch materials, but the practical takeaway for builders is simpler: the stronger tier is not being offered with the same access path as the base model.