Muse Wonder
Meta's first frontier model, built for long-running agent tasks
Context window
1M
Input / 1M tokens
$4.00
Output / 1M tokens
$20.00
Provider
Meta
Prices and context from TechCrunch, VentureBeat and Artificial Analysis; Meta does not list them on its own pages. Cached input $0.20 per 1M. The stronger Muse Wonder Full is available only to Muse Code subscribers, with no API. · Data verified 2026-10-15
Muse Wonder (October 12, 2026) is the first frontier-class model from Meta Superintelligence Labs, built for long-running personal and professional agent work with what Meta calls Active Memory and Active Focus. It accepts text, image, audio, video and PDF input. It is available through the Meta Model API and Muse Code; a stronger variant, Muse Wonder Full, is limited to Muse Code subscribers. Unlike Meta's earlier Llama models, it is proprietary.
Capability index
Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.
How to access it
Meta Model API and Muse Code; also listed on OpenRouter as meta/muse-wonder.
Strengths
- ✓Frontier-level intelligence: 52 on the Artificial Analysis Intelligence Index (Full: 56)
- ✓Strong coding: 78.8 on DeepSWE 1.1 (third-party reported)
- ✓Natively multimodal input including audio and video
- ✓Priced on par with Claude Opus 5.5
Best for developers who...
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| Artificial Analysis Intelligence Index | 52 | Third-party; Muse Wonder Full 56, Claude Opus 5.5 58, GPT-6 Astra 53 |
| DeepSWE 1.1 | 78.8 | Third-party reported; Full 79.0 |
| SWE-Marathon | 53.1 | Third-party reported; Full 57.5 |
Source: Artificial Analysis
Compare Muse Wonder
Compare Muse Wonder with any other model
Build a comparison →All model comparisons →