Meta announced on October 12, 2026 the launch of Muse Wonder and Muse Wonder Full, described by the company as its most capable models to date and the first frontier models from Meta Superintelligence Labs.
The models were designed for personal and professional agentic tasks, with emphasis on Active Memory — the ability to carry relevant context across conversations — and Active Focus — persistence on complex goals, trying alternative approaches and recovering from errors.
Muse Wonder is immediately available via the Meta Model API (priced at $4 per million input tokens and $20 output, with cached input at $0.20) and Muse Code. Muse Wonder Full, the more advanced version, is restricted to Muse Code subscribers on the Power and Maximum plans.
According to Meta, the models achieve state-of-the-art performance in domains such as law, finance, health, and coding. Independent evaluations from Artificial Analysis place Muse Wonder Full at 56 points on the Intelligence Index, tying with Claude Sonnet 5.5 and behind only Claude Opus 5.5.
Mark Zuckerberg and Alexandr Wang of Meta Superintelligence Labs highlighted the advance after about 14 months of restructuring the company's AI efforts, trained at the gigawatt-scale Prometheus data center.
Muse Wonder will soon power experiences in the personal agent Muse, starting with subscribers.
Sources
Image credit: Meta / Mark Zuckerberg
Transparency: This content was created, edited, or reviewed with the assistance of artificial intelligence. The information was cross-checked with public posts on X and sources available on the internet. Consult the original sources to verify the full context.
By GeekikiBot
