Meta AIopen-weight
Released: 2026-11-09Artificial AnalysisMuse Ember
Muse Ember is the smallest and most efficient model from Meta that’s designed for complex reasoning execution and the work agents repeat all day: tool calls, coding agent subtasks, routing,...
Intelligence Index
36.0/ 100
Artificial Analysis Intelligence Index (independent composite)Throughput Speed
425tok/sec
Output streaming throughputTime to First Answer
5.2s
Time to first answer token, including any thinkingCost to Run the Index
0.14USD
Artificial Analysis cost of the full Index runMeasured results · Artificial Analysis
Independent evaluations of Muse Ember (Max configuration). Values are reported as published; nothing here is estimated or derived. Snapshot 2026-11-15.
- Intelligence Index
- 36.0
- Cost to run the Index
- $0.14
- Time per task
- 3.3 min
- Omniscience index
- 7.4
- Reasoning model
- Yes
- Cached-input price
- $0.004 / 1M
Reasoning & knowledge
- AA-Omniscience Accuracy21.1%
- CritPt v2 (Hard)39.0%
- HLE-Diamond25.1%
Coding & terminal agents
Agentic work
- AA-Briefcase v1.1 (Elo)1520 Elo
- AutomationBench-AA68.1%
Long context & documents
- GDP.pdf22.8%
- MLCR-AA (medical long context)18.3%
Health & safety
Reasoning-effort variants evaluated
| Variant | Intelligence Index | Cost to run Index | Output speed (tok/s) |
|---|---|---|---|
| Max | 36.0 | $0.14 | 425.4 |
| Xhigh | 34.1 | $0.12 | 418.9 |
Headline Evaluation Breakdown
Six independently measured evaluations (Artificial Analysis)Reasoning · CritPt v2 (Hard)39.0%
Coding & agents · Terminal-Bench 4.040.9%
Sourced Benchmark Evaluations (10)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| Terminal-Bench 4.0 | Active | 40.9% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench-Science 0.1 | Active | 20% | 93 | 2026-11-15 | independent | Source ↗ |
| CritPt v2 (Hard) | Active | 39% | 93 | 2026-11-15 | independent | Source ↗ |
| HLE-Diamond | Active | 25.1% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-Omniscience (Accuracy) | Active | 21.1% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-Omniscience (Non-Hallucination) | Active | 82.6% | 93 | 2026-11-15 | independent | Source ↗ |
| FrontierCode-AA 2.0 | Active | 27.2% | 93 | 2026-11-15 | independent | Source ↗ |
| GDP.pdf | Active | 22.8% | 93 | 2026-11-15 | independent | Source ↗ |
| MLCR-AA (Medical Long Context Reasoning) | Active | 18.3% | 93 | 2026-11-15 | independent | Source ↗ |
| AutomationBench-AA | Active | 68.1% | 93 | 2026-11-15 | independent | Source ↗ |
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run meta-muse-ember
# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "meta-muse-ember",
"messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
}'Evaluation & Sourcing Notes
Scores listed for Muse Ember represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.