Stories
Meta Unveils Muse Wonder AI Models Featuring Active Memory and Focus
F
Firstpost•12-10-2026, 23:58
Meta Unveils Muse Wonder AI Models Featuring Active Memory and Focus
- •Meta introduced the Muse Wonder series, featuring Muse Wonder and Muse Wonder Full, designed for advanced personal and professional agentic tasks.
- •The new models introduce Active Memory to retain user context across conversations and Active Focus for persistent task execution.
- •Meta claims these models outperform competitors like OpenAI and Anthropic on benchmarks for legal analysis and software development.
- •Muse Wonder and Muse Wonder Full include robust safety safeguards against CBRN risks and cyber misuse while delivering frontier coding performance.
Key Questions Answered
Meta introduced the Muse Wonder series, which includes Muse Wonder and Muse Wonder Full. These models are designed to provide advanced performance for personal and professional agentic tasks by incorporating Active Memory and Active Focus features.
Active Memory helps AI agents retain important user context across conversations and retrieve relevant details when needed. Active Focus allows the agent to work persistently on a task, explore different approaches, update users on progress, and request clarification before taking major steps.
Meta stated that its new models outperformed OpenAI's GPT-6 Astra and Anthropic's Claude Fable 5.1 and Claude Opus 5.5 on select specialised benchmarks like legal document analysis and software development.
