LATEST MODEL
View in the Herd

Muse Wonder

Meta ⚔️ The Knight Released October 2026

Meta's first frontier-class model, closed-weight and built for personal and professional agents, with Active Memory across conversations and Active Focus on long goals; its larger sibling, Muse Wonder Full, is reserved for paid Muse Code subscribers

Muse Wonder

Meta • October 2026

Latest

Training Data

Not disclosed

Muse Wonder

October 2026

Parameters

Not disclosed

Training Method

Pretraining and reinforcement learning at Meta's gigawatt-scale Prometheus data center, on larger base models than Muse Spark 1.3

Context Window

1,048,576 tokens

Knowledge Cutoff

Not disclosed

Key Features

Active Memory Across Conversations • Active Focus on Long Goals • Wonder Full for Muse Code Subscribers • Closed Weights

Capabilities

Intelligence: Outstanding (AA Index 52; Wonder Full 56)

Long-Running Coding: Outstanding (Wonder Full 57.5 SWE-Marathon)

Agentic: Outstanding

What's New in This Version

The first Meta model at the frontier: Wonder Full ties Claude Sonnet 5.5 on the Artificial Analysis Intelligence Index, built on larger base models than Muse Spark 1.3; $4/$20 per 1M tokens through the Meta Model API (muse-wonder-1.0), with no downloadable weights for either size

Meta's first frontier-class model, closed-weight and built for personal and professional agents, with Active Memory across conversations and Active Focus on long goals; its larger sibling, Muse Wonder Full, is reserved for paid Muse Code subscribers

What's New in This Version

The first Meta model at the frontier: Wonder Full ties Claude Sonnet 5.5 on the Artificial Analysis Intelligence Index, built on larger base models than Muse Spark 1.3; $4/$20 per 1M tokens through the Meta Model API (muse-wonder-1.0), with no downloadable weights for either size

Technical Specifications

Parameters Not disclosed
Context Window 1,048,576 tokens
Training Method Pretraining and reinforcement learning at Meta's gigawatt-scale Prometheus data center, on larger base models than Muse Spark 1.3
Knowledge Cutoff Not disclosed
Training Data Not disclosed

Key Features

Active Memory Across Conversations Active Focus on Long Goals Wonder Full for Muse Code Subscribers Closed Weights

Capabilities

Intelligence: Outstanding (AA Index 52; Wonder Full 56)
Long-Running Coding: Outstanding (Wonder Full 57.5 SWE-Marathon)
Agentic: Outstanding
Theme
Language
Support
© funclosure 2025