← Today’s paper

From the paper of Tuesday, October 13, 2026

Meta releases Muse Wonder and Muse Wonder Full, its first frontier AI models

No. 4·366 words·2 min

Meta released two AI models Monday, Muse Wonder and Muse Wonder Full, the first frontier models from its Superintelligence Labs division. Wonder Full is available only to Muse Code subscribers and will not be served through an API. Muse Wonder is sold through the Meta Model API at $4 per million input tokens and $20 per million output tokens, roughly what GPT-6 Sol or Claude Opus 5.5 cost, according to TechCrunch. Both models will also be used to power Meta's Muse agent.

Meta says the pair matches the industry's best. That claim rests on the company's own testing, presented in the models' card: Meta reports benchmark scores comparable to Claude Fable 5.1 and GPT-6 Astra. The Verge reports Meta points to internal tests as evidence the models can rival leading offerings from OpenAI and Anthropic, including GPT-6 Astra, Claude Opus 5.5 and Claude Fable 5.1. Neither source says how the Wonders compare to Google's Gemini 4 Argon, whose capabilities are still coming into focus.

The pitch is memory. Meta placed particular emphasis on what it calls "active memory" — the models' ability to retain and synthesize information a user gives them over time. On MemoryBench, a benchmark that tests continual learning and memory construction, Meta's testing puts Muse Wonder above Fable, Opus and Astra.

"Together, they mark the next step on our scaling ladder and a major leap toward personal superintelligence," Meta's blog post says. Research scientist Kavya Kopparapu, in a video accompanying the launch, said: "We put a tremendous amount of effort into making sure your model can remember what's important to you over days, weeks, months, and beyond."

Meta says it eventually plans to power its Muse agent with Wonder, starting with subscribers, and to develop specialized models it "can bring to everyone using Muse." It gave no specifics and no timeline for either.

TechCrunch reports the models were trained at Meta's Prometheus data center in central Ohio, which came online earlier this year. They are the first frontier release since Scale AI founder Alexandr Wang restructured Meta's AI efforts in August of last year. Since then the company's releases have been the lightweight Muse Spark, the Muse agent, and image model Muse Image.

Discuss on the board