Muse Ember
muse-ember
Muse Ember is Meta's smallest and most efficient reasoning model, built for the work agents repeat all day: tool calls, coding subtasks, routing, extraction, and long-document parsing. It pairs with Muse Wonder or Muse Spark, which plan while Muse Ember handles the volume.
ReasoningTool CallingAttachments
Context Window
1.0M
Max Output
524K
Temperature
Yes
Open Weights
No
Knowledge Cutoff
N/A
Released
2026-11-09
Last Updated
2026-11-09
Modalities
Input:TextImageVideoPDFAudio
→Output:Text
Benchmarks
| Benchmark | Score | Metric | Harness |
|---|---|---|---|
| AutomationBench | 37.8 | strict pass@1 | - |
| Agents' Last Exam | 61.6 | score | Meta internal evaluation framework |
| Terminal-Bench | 40.9 | accuracy | mini-swe-agent |
| FrontierCode | 27.2 | score | - |
| GDP.pdf | 22.8 | score | - |
| OSWorld | 69.5 | mean partial score | - |
| OSWorld | 33.8 | binary completion rate | - |
| PerceptionBench | 80.6 | score | Meta internal evaluation framework |
| Artificial Analysis Intelligence Index | 36 | index score | - |
Links
Available Providers (2)
| Provider | Input /1M | Output /1M | Cache Read /1M | Cache Write /1M | Reasoning | Status |
|---|---|---|---|---|---|---|
| Rp 1.817 | Rp 9.082 | Rp 73 | — | effort | — | |
| Rp 1.817 | Rp 9.082 | Rp 73 | — | effort | — |