Meta has released Muse Wonder and Muse Wonder Full, which the company describes as its most capable models to date for personal and professional agentic tasks.
According to Meta’s announcement on October 12, 2026, the models are designed to handle complex, multi-step work such as navigating insurance claims, interpreting legal documents, managing finance tasks, and building applications. They include two new capabilities: Active Memory, which carries relevant context across conversations, and Active Focus, which keeps the model working toward a stated goal, recovers from errors, and checks in before taking significant actions.
Muse Wonder is available now through the Meta Model API and Muse Code. Muse Wonder Full is limited to higher-tier Muse Code subscribers. Both support a one-million-token context window and accept text, image, video and audio as input.
Meta says the models achieve competitive results on independent benchmarks. On the Artificial Analysis Intelligence Index, Muse Wonder Full scored 56, tying one of Anthropic’s models and trailing Claude Opus 5.5. The company also published results showing strength in long-horizon coding tasks and certain professional domains compared with GPT-6 Astra and other frontier models.
The release continues Meta’s push into frontier AI after the earlier Muse Spark models. Muse Wonder is priced at $4 per million input tokens and $20 per million output tokens on the API. The company said the models will eventually power experiences in its Muse assistant, starting with subscribers.
Source: Meta Newsroom and research blog.
