Meta: Muse Wonder
Muse Wonder is a frontier multimodal reasoning model from Meta for personal and professional agents, long-horizon coding, and computer use. It combines Active Memory, which carries relevant context across conversations, with Active Focus, which keeps it working through blockers toward a goal.
Sources
Same model, different sources. Our first-party rate is listed first, then router and aggregator rows stay separate so you can see the spread instead of a single blended number.
| Source | Provider | Input | Output | Cache read | Context |
|---|---|---|---|---|---|
| tokenstatcheapestmeta/muse-wonder-1.0 | Meta | $4 | $20 | $0.2 | 1.0M |
| models.devmeta/muse-wonder-1.0 | Cloudflare | $4 | $20 | $0.2 | 1.0M |
| models.devmeta/muse-wonder-1.0 | CommandCode | $4 | $20 | $0.2 | 1.0M |
| models.devmeta/muse-wonder-1.0 | Eden AI | $4 | $20 | $0.2 | 1.0M |
| models.devmeta/muse-wonder-1.0 | LLM Gateway | $4 | $20 | $0.2 | 1.0M |
| models.devmuse-wonder-1.0 | Meta | $4 | $20 | $0.2 | 1.0M |
| models.devmeta/muse-wonder-1.0 | Vercel AI Gateway | $4 | $20 | $0.2 | 1.0M |
| models.devmeta/muse-wonder | Kilo Gateway | $4 | $20 | $0.2 | 1.0M |
| models.devmeta/muse-wonder | NanoGPT | $4 | $20 | $0.2 | 1.0M |
| OpenRoutermeta/muse-wonder | OpenRouter | $4 | $20 | $0.2 | 1.0M |
| models.devmuse-wonder-1.0 | DevPass (LLM Gateway) | $4 | $20 | $0.2 | 1.0M |
| models.devmuse-wonder-1.0 | OpenCode | $4 | $20 | $0.2 | 1.0M |
| models.devmuse-wonder-1.0 | Requesty | $4 | $20 | $0.2 | 1.0M |
| LiteLLMmeta/muse-wonder | N/A | $4 | $20 | $0.2 | N/A |
| LiteLLMmuse-wonder-1.0 | N/A | $4 | $20 | $0.2 | N/A |
| LiteLLMmuse-wonder | N/A | $4 | $20 | $0.2 | N/A |
| models.devmuse-wonder | Venice AI | $5 | $25 | $0.25 | 1.0M |
| LiteLLMkyma/muse-wonder | N/A | $5.40 | $27 | $0.54 | N/A |
Capabilities
- Reasoning
- Tool calling
- Structured output
- Attachments
- In: text
- In: image
- In: video
- In: pdf
- In: audio
- In: file
- Out: text
Scores
Sparse coverage on purpose. Missing a score means we have not linked one yet, not that the model is unranked. Hover or tap the i next to any score for its source. Full boards on /benchmarks.
- Terminal-Bench 4.052i
- DeepSWE v1.178.8i
- AutomationBench (Public)39.1i
- HealthBench Professional (length-adjusted)60.5i
- OSWorld 2.1 (partial)72.6i
- GDP.pdf30.8i
- Vals Finance Agent v256.6i
- Personal Superintelligence Index (Meta)54.3i
- Harvey LAB-AA v1.110i
- JobBench70.5i
- OSWorld 2.1 (108 tasks, binary)34.3i
- MemoryBench (Meta)85.6i
- SWE-Marathon53.1i
- ProgramBench (at least 95% tests)50i
- AA Intelligence Index v4.3 (Meta-reported)52i
- FocusBench (Meta)44.9i
- Personal information violation rate (lower is better)3.4i
- Unauthorized code execution ASR (lower is better)10.5i
- BioTIER refusal (launch chart)98i
- Cyber misuse refusal (Meta)93.7i
- Cyber defense compliance (Meta)94.5i
Arena rows are community-vote Elo from Arena AI, Artificial Analysis indices arrive via OpenRouter, and launch-eval figures are collected from vendor announcements, analyst writeups, and community benchmarks. Full boards and credits on /benchmarks.
See what this model costs you
The app reads the sessions your tools already wrote and prices this model's tokens at the published API rate. Free on your computer.