Muse Wonder
muse
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.
ReasoningTool CallingAttachments
Context Window
1.0M
Max Output
131K
Temperature
Yes
Open Weights
No
Knowledge Cutoff
N/A
Released
2026-10-12
Last Updated
2026-10-12
Modalities
Input:TextImageVideoPDFAudio
→Output:Text
Benchmarks
| Benchmark | Score | Metric | Harness |
|---|---|---|---|
| Harvey Legal Agent Benchmark | 10 | all-pass rate | - |
| Finance Agent | 56.6 | accuracy | - |
| HealthBench Professional | 60.5 | length-normalized weighted rubric score | - |
| AutomationBench | 39.1 | strict pass@1 | - |
| GDP.pdf | 30.8 | score | - |
| JobBench | 70.5 | mean reward | - |
| OSWorld | 72.6 | mean partial score | - |
| OSWorld | 34.3 | binary completion rate | - |
| Terminal-Bench | 52 | accuracy | - |
| DeepSWE | 78.8 | pass@1 | - |
| SWE-Marathon | 53.1 | pass@1 | - |
| ProgramBench | 50 | share of rollouts passing at least 95% of tests | mini-swe-agent |
| Artificial Analysis Intelligence Index | 52 | index score | - |
Links
Available Providers (5)
| Provider | Input /1M | Output /1M | Cache Read /1M | Cache Write /1M | Reasoning | Status |
|---|---|---|---|---|---|---|
 Eden AI | Rp 72.516 | Rp 362.579 | Rp 3.626 | — | effort | — |
 CommandCode | Rp 72.516 | Rp 362.579 | Rp 3.626 | — | effort | — |
 LLM Gateway | Rp 72.516 | Rp 362.579 | Rp 3.626 | — | effort | — |
 Cloudflare AI Gateway | Rp 72.516 | Rp 362.579 | Rp 3.626 | — | effort | — |
 Vercel AI Gateway | Rp 72.516 | Rp 362.579 | Rp 3.626 | — | effort | — |