Meta's Muse Wonder is a frontier reasoning model designed for complex personal and professional agentic tasks. It accepts text, images, audio, video, and PDF documents across a one-million-token context window, supports tool calling and structured output, and always reasons before answering.
Specification
most-agreed values
Context
1M
Max output
1M
Released
2026-10-12
Knowledge cutoff
—
Retires
—
Open weights
no
Input
text, image, video, audio, pdf
Output
text
Price
US dollars per million tokens · most-agreed
Input
$4.00
Output
$20.00
Cache read
$0.20
Quality
6 benchmarks · 1 source
Composite
92nd
percentile of 344 scored models
Rank
30 / 344
±0.08 sd
Evidence
6 × 1
one source, or fewer than 3 benchmarks
Effort range
—
one configuration only
reasoning87th2/4 bench
math92nd3/7 bench
Every score
one row per source and configuration — nothing averaged away
| Benchmark | Score | Configuration | Source | Run |
|---|---|---|---|---|
| Mystery Game Puzzles | 51.0 ±5.0 | max | Epoch AI | 2026-10-14 |
| FrontierMath-Tiers-1-3-v2 | 82.1 ±2.3 | max | Epoch AI | 2026-10-12 |
| OTIS Mock AIME 2024-2025 | 99.4 ±0.4 | max | Epoch AI | 2026-10-12 |
| FrontierMath-Tier-4-v2 | 70.7 ±7.2 | max | Epoch AI | 2026-10-12 |
| Furniture Assembly | 54.2 ±6.4 | max | Epoch AI | 2026-10-15 |
| EBR-bench | 31.4 ±4.2 | max | Epoch AI | 2026-10-12 |
The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.
Available from 5 hosts
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| kilo | $4.00 | $20.00 | $0.20 | — | 1M | 944K | — |
| nano-gpt | $4.00 | $20.00 | $0.20 | — | 1M | 1M | — |
| openrouter | $4.00 | $20.00 | $0.20 | — | 1M | 944K | — |
| venice | $5.00 | $25.00 | $0.25 | — | 1M | 131K | — |
| kyma · kyma | $5.40 | $27.00 | — | — | 1M | 944K | — |
More from Meta
most-hosted first