agents

Meta's Muse Wonder models are built for agents that remember and keep going

Muse Wonder and Muse Wonder Full add Active Memory and Active Focus, two features aimed at long-running personal and professional agent tasks.

Written by Mara at Sig · Oct 18, 2026

Meta introduced Muse Wonder and Muse Wonder Full on October 12, calling them its most capable models to date. Meta pitches them for personal and professional agent work such as insurance claims, lease review and personal finances, as well as coding.

Two features stand out. Active Memory keeps relevant details across conversations and pulls them into the current task. Active Focus is meant to keep an agent working toward a goal: pushing through blockers, trying alternatives, recovering from errors, and asking for confirmation before consequential actions.

Muse Wonder is available through the Meta Model API and Muse Code, Meta's coding product. Muse Wonder Full is available to Muse Code subscribers. According to Aiholics, Meta's announcement did not include benchmark scores or pricing, and OpenRouter's catalogue lists Muse Wonder at $4 per million input tokens and $20 per million output tokens.

Why it matters

Memory across sessions and confirm-before-acting usually live in your own agent code. If the model handles them, that code can get smaller. Check the confirmation behavior on your own risky actions before relying on it.