- AutomationBench-AA, Muse Ember max reasoning 37.8
- Agents’ Last Exam, Muse Ember max reasoning 61.6
- Personal Superintelligence Index, Muse Ember max reasoning 46.8
- Terminal-Bench 4.0, best published harness, Muse Ember max reasoning 40.9
- AA Coding Agent Index v2.0, Muse Ember max reasoning 46
- FrontierCode 2.0-AA, Muse Ember max reasoning 27.2
- GDP.pdf-AA, Muse Ember max reasoning 22.8
- OSWorld 2.1, partial-credit reward, Muse Ember max reasoning 69.5
- PerceptionBench with tools, Muse Ember max reasoning 80.6
Muse Ember
Muse Ember ranks first on 2 of the 40 benchmarks with five or more models scored.
41
Benchmarks with a score
2
Ranked first, out of five or more models
5
In the top three
Meta Nov 2026 Muse Ember
Closed weights Small
No Epoch Capabilities Index score yet
- GDP.pdf 49.2% #7 /19
- OSWorld 2.1 69.5% #7 /7
- Legal Research Bench 49% #12 /93
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BullshitBench V2 | 62.6% | #33 of 129 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SimpleBench | 72.7% | #21 of 106 | Official leaderboard |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DesignArena Image to HTML | 1,273 | #2 of 51 | Official leaderboard | |
| DesignArena Website | 1,331 | #7 of 151 | Official leaderboard | |
| DesignArena Code | 1,348 | #10 of 143 | Official leaderboard | |
| DesignArena 3D Design | 1,374 | #12 of 136 | Official leaderboard | |
| DesignArena Data Visualization | 1,317 | #12 of 118 | Official leaderboard | |
| DesignArena Game Dev | 1,332 | #15 of 143 | Official leaderboard | |
| DesignArena UI Component | 1,312 | #19 of 139 | Official leaderboard | |
| APEX-SWE | 52.9% | #11 of 65 | Official leaderboard | |
| Vibe Code Bench | 83.7% | #24 of 120 | Official leaderboard | |
| Code Arena | 1,609 | #31 of 128 | Official leaderboard | |
| FrontierCode 2.0 | 27.2% | #3 of 6 | Lab model card | |
| SWE Atlas v1.1 | 56% | #20 of 28 | Official leaderboard | |
| DAXBench v2 | 59.4% | #44 of 47 | Official leaderboard | |
| Artificial Analysis Coding Agent Index v2 | 46.0 | #1 of 1 | Lab model card |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Agent Arena | 1.67% | #25 of 56 | Official leaderboard | |
| Terminal-Bench 4.1 | 30.6% | #20 of 43 | Official leaderboard | |
| AutomationBench | 37.8% | #25 of 47 | Lab model card | |
| LLM Chess | -38 | #69 of 123 | Official leaderboard | |
| Terminal-Bench 4.0 | 40.9% | #16 of 28 | Lab model card | |
| RuneBench | 4.51 | #44 of 76 | Official leaderboard | |
| OSWorld 2.1 | 69.5% | #7 of 7 | Lab model card |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Agents' Last Exam | 61.6% | #1 of 26 | Lab model card | |
| Harvey LAB v1.1 (HLAB-Vals-1.1) | 25% | #3 of 38 | Official leaderboard | |
| Vals Legal Index | 75.7% | #4 of 38 | Official leaderboard | |
| Legal Research Bench | 49% | #12 of 93 | Official leaderboard | |
| APEX-1 | 70.4% | #10 of 71 | Official leaderboard | |
| Tax Agent Bench | 38.2% | #15 of 84 | Official leaderboard | |
| APEX-Agents 1.1 | 75.7% | #16 of 65 | Official leaderboard | |
| APEX-Agents | 64.3% | #16 of 61 | Collected by Epoch AI, source not named | |
| Vals Finance Index | 60.9% | #12 of 40 | Official leaderboard | |
| Excel Modeling Benchmark v1.1 | 72.2% | #13 of 42 | Official leaderboard | |
| FoodTruck Bench | $34,639 | #19 of 59 | Official leaderboard | |
| Vals Index v2.2 | 60.5% | #14 of 37 | Official leaderboard | |
| Vals Finance Agent v2.1 | 63.8% | #16 of 40 | Official leaderboard | |
| DeepFinance | 36.5% | #11 of 23 | Official leaderboard | |
| YC-Bench | $1,018,338 | #33 of 59 | Official leaderboard | |
| EEBench | 42.8% | #22 of 30 | Official leaderboard |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PerceptionBench (Moonshot) | 80.6% | #1 of 5 | Lab model card | |
| GDP.pdf | 49.2% | #7 of 19 | Official leaderboard |
Its lab's number next to a leaderboard
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.
Its system card
Every result in the longer documents the lab published with it.
Muse Ember Evaluation Report Nov 9, 2026
Capabilities
- AutomationBench 37.8
- Agents' Last Exam 61.6
- Personal Superintelligence Index (Meta) 46.8
- Terminal-Bench 4.0 40.9
- Artificial Analysis Coding Agent Index v2 46
- FrontierCode 2.0 27.2
- GDP.pdf 22.8
- OSWorld 2.1 69.5
- PerceptionBench (Moonshot) 80.6