Meta takes aim at the AI frontier: Muse Wonder is its first frontier model
Meta's Superintelligence Lab has released Muse Wonder and Muse Wonder Full. The models are built to retain information across conversations and to rival competitors' top models.
Archive image · Photo: LPS.1 / Wikimedia Commons, CC0 (cropped and resized)
Meta on Monday released Muse Wonder and Muse Wonder Full, its first frontier models, meaning AI systems meant to compete with the industry's strongest models. The company calls them "our most capable models to date." They are larger and more capable than the lightweight Muse Spark and are aimed mainly at coding and complex agentic tasks such as insurance claims, leases and personal finances.
The headline feature is "active memory": the models are designed to retain important information across conversations. A second capability, Active Focus, is meant to keep the AI working toward a goal when a task gets hard, recover from errors and check in before major actions. According to Meta's own testing, as reported by TechCrunch, the models post benchmark scores comparable to GPT-6 Astra and Fable 5.1. They were trained at Meta's Prometheus data center in Ohio.
Muse Wonder is available to developers through the Meta Model API, priced at $4 per million input tokens and $20 per million output tokens, according to TechCrunch. The larger Muse Wonder Full is limited to Muse Code subscribers. It is Meta's first release of this kind since Alexandr Wang restructured its AI efforts in August 2025. Meta says a full safety report will follow soon.