Open-weight Muse model that Meta calls the most capable US open-weight model to date. Natively multimodal across text, image, video, and audio, built for coding and long-horizon agentic work, with weights on Hugging Face and an MLX build tuned for Mac Studio.
SI v3 uses weighted percentages from one Artificial Analysis configuration. Ranking requires all five benchmarks. Partial scores average only available weights. Scores are not comparable to SI v2 or a probability of reaching singularity.
Compared with models released within six months of it.
| Benchmark | Score | Rank |
|---|---|---|
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 43.9% | #25 / 100 |
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 89% | #46 / 96 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisMuse Spark Open (Xhigh) | 43.9% |
MMMU-Pro Artificial AnalysisMuse Spark Open (Xhigh) | 80.9% |
Terminal-Bench 2.1 Artificial AnalysisMuse Spark Open (Xhigh) | 76.4% |
Terminal-Bench 4.0 Artificial AnalysisMuse Spark Open (Xhigh) | 8.1% |
SciCode Artificial AnalysisMuse Spark Open (Xhigh) | 58.4% |
AA-LCR Artificial AnalysisMuse Spark Open (Xhigh) | 78.3% |
CritPt Artificial AnalysisMuse Spark Open (Xhigh) | 14.9% |