Meta Muse Wonder: Meta's first frontier AI model, explained
On October 12, 2026, Meta unveiled Meta Muse Wonder and Muse Wonder Full, the first frontier models from its Superintelligence Labs. They focus on long-term memory and agents that finish complex tasks: here is what they do, what they cost and how to use them.
Leggi questo articolo in italiano
What is Meta Muse Wonder?
Meta Muse Wonder is a family of two large language models that Meta introduced on October 12, 2026: Muse Wonder and the more powerful Muse Wonder Full. In its official announcement, Meta describes them as its most capable models, built for agents: systems that don't just answer but carry out multi-step tasks on the user's behalf.
It is a milestone for the company. As TechCrunch notes, it is Meta's first frontier model since it reorganized its AI teams under Alexandr Wang in August 2025. Before this came Muse Spark, a lightweight, fast model launched in April 2026, and the Muse agent.
Meta says it trained the Wonder series at its Prometheus data center in Ohio, combining pretraining and reinforcement learning. If you want to find your way around generative AI tools, our AI guides offer an overview (in Italian).
What's new: Active Memory and Active Focus
Meta built the launch around two capabilities, which are also the most useful in practice.
- Active Memory: the model keeps important context about you across conversations and retrieves the details relevant to the task at hand. Meta sees this as central for personal assistants working on long projects.
- Active Focus: the model keeps pushing toward hard goals, tries alternative approaches when stuck, adapts to new information, asks for clarification and checks in before important actions.
Meta's demos show very concrete use cases: a small game built from image inputs, an insurance claim, running-route planning, personalized gift ideas and wedding seating charts.
Meta Muse Wonder: price and availability
According to Meta, Muse Wonder is available now through the Meta Model API and Muse Code, while Muse Wonder Full is reserved for Muse Code subscribers. Meta adds that Muse Wonder will soon power the Muse app as well, rolling out to subscribers first. The announcement does not list supported countries, so check availability on Meta's own page.
| Model | Where to use it | API price |
|---|---|---|
| Muse Wonder | Meta Model API, Muse Code; coming to the Muse app | $4 per million input tokens, $20 per million output tokens (source: TechCrunch) |
| Muse Wonder Full | Muse Code subscribers only | Not available via API |
As a worked example: a job that sends the model 1 million tokens and gets 200,000 back would cost $4 + $4, about $8, at the prices reported by TechCrunch.
How powerful is it compared with rivals?
Meta claims the Wonder series delivers state-of-the-art results on benchmarks for personal and professional agents, memory and coding, citing among others Harvey's Legal Agent Benchmark, FocusBench and SWE-Marathon (long-running software projects). According to TechCrunch, Meta's own testing puts both models on par with Fable 5.1 and GPT-6 Astra, and ahead of rivals on MemoryBench, the memory benchmark.
One thing to keep in mind: these are the company's own numbers. Meta says a full safety report is coming soon; for a solid verdict, wait for independent evaluations and, above all, test the model on your own real tasks.
On safety, Meta calls Muse Wonder its "most aligned" model to date and says it has stronger safeguards against chemical, biological, radiological, nuclear and cyber misuse.
What it means for you and how to use it today
For creators and solo professionals, the real news is not one more benchmark point but memory. An assistant that remembers your style, clients and projects saves you from re-explaining everything each time. Here is how to put it to the test:
- Build a "work profile". In your first chat, describe your tone of voice, audience, formats (posts, short videos, newsletters) and fixed rules. In later sessions, check whether the model applies them without being reminded.
- Give it a multi-stage project. For example: a one-month content plan, then drafts, then a review pass. That is exactly the ground Meta says Active Focus was built for.
- Compare on your real work. Give the same task to Muse Wonder and to the assistant you already use, and judge the output, not the leaderboard.
- Keep sensitive data separate. Don't upload confidential information until you understand how it is stored.
If you want to turn these tools into professional videos, images and avatars, The AI Creator System course walks you through it step by step. For hands-on help with a business project, see our done-for-you services.
Frequently asked questions
What is Meta Muse Wonder?
It is Meta's first frontier AI model, unveiled on October 12, 2026 together with Muse Wonder Full. It is designed for assistants and agents that remember the user's context and complete long tasks.
How much does Muse Wonder cost?
According to TechCrunch, the Meta Model API charges $4 per million input tokens and $20 per million output tokens. Muse Wonder Full is not offered through the API, only to Muse Code subscribers.
What is the difference between Muse Wonder and Muse Wonder Full?
Muse Wonder Full is the most powerful model in the series and is limited to Muse Code subscribers. Muse Wonder is also available to developers through the API.
Is Muse Wonder available in my country?
Meta's announcement does not list supported countries. The most reliable way to find out is to check Meta's official page and the Meta Model API dashboard from your own account.
