CONFIRMED: Meta announced the models and their availability on 12 October 2026. Benchmark results and rankings are COMPANY CLAIMS. A comparison with Anthropic’s Claude models is REPORTED by Wired and has not been independently verified.
Meta has introduced Muse Wonder and Muse Wonder Full, which it calls “our most capable models to date, built to deliver state-of-the-art performance on personal and professional agentic tasks — from navigating an insurance claim to making sense of a lease to building a working app” [1].
Agentic tasks are jobs where an AI model takes a series of steps on its own (using apps, browsing or writing code) to reach a goal, rather than just answering one question.
Who can use them, and when
- “Muse Wonder is available today through the Meta Model API and Muse Code, and Muse Wonder Full is available today for Muse Code subscribers” [2].
- “Muse Wonder will soon power experiences in Muse, and will roll out first to subscribers” [2].
Meta did not give prices, model sizes or benchmark scores in the text of its announcements.
How they were built
Meta said the models build “on the groundwork we laid with Muse Spark” [1]. The research blog says: “Trained at our gigawatt-scale Prometheus data center, Muse Wonder and Muse Wonder Full bring together the work of teams across Meta” [2].
Active Memory and Active Focus
Meta describes two headline features [1]:
- Active Memory: “the ability to carry what matters to you across conversations and bring back the details that are relevant to the task at hand.”
- Active Focus: “the ability to keep working toward your goal when a task gets hard, try other approaches, recover from errors, and follow through, asking for clarification when needed and checking in before following through on anything important.”
What Meta says about performance (COMPANY CLAIM)
Meta says “The Muse Wonder series achieves state-of-the-art results across key benchmarks for personal and professional agents, memory, and coding” [2]. Its evaluations include Harvey’s Legal Agent Benchmark (LAB), which “requires models to work with source documents and follow detailed instructions to produce legal work products”. Meta says it measures focus “through FocusBench and sustained execution through SWE-Marathon, which tests work on long-running software projects” [2].
A benchmark is a standard set of test tasks used to compare AI models. Companies often choose and run their own, so results are not independent.
Wired reports that “Meta says the largest version of the new model outperforms Anthropic’s second-best model, Claude Fable 5.1, and ranks just below its best, Claude Opus 5.5, on the Artificial Analysis Intelligence Index” [3]. TSN has not found this ranking in Meta’s own announcement text or on Artificial Analysis.
Safety
Meta says it trained the models “to avoid sharing private information without permission” and to “turn down requests to assist in performing actually harmful acts” [1]. It says the models are released in line with its Meta Superintelligence Scaling Framework, and that “A full report with additional safety evaluations will be available soon” [1][2].
Sources
- Meta Newsroom, “Introducing Muse Wonder and Muse Wonder Full: Frontier Models for Agents”, 12 October 2026. https://about.fb.com/news/2026/10/introducing-muse-wonder-models/
- Meta AI Research, “Muse Wonder”, 12 October 2026. https://research.meta.ai/blog/muse-wonder
- Wired (Will Knight), “Muse Wonder Puts Meta Back at the AI Frontier”, 12 October 2026. https://www.wired.com/story/muse-wonder-puts-meta-back-at-the-ai-frontier/
![Pinterest](https://pinterest.com/pin/create/button/?url=https://tsnmedia.org/meta-muse-wonder-full-agentic-models-meta-model-api-muse-code/&media=https://tsnmedia.org/wp-content/uploads/2026/10/meta-muse-wonder-full-agentic-models-meta-model-api-muse-code.jpg&description=Meta launches Muse Wonder and Muse Wonder Full, "our most capable models to date", for agentic tasks, via the Meta Model API and Muse Code from today.){kind=link}