Meta's first proprietary model from Superintelligence Labs, non-open-weight
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.
| Benchmark | Score | Rank |
|---|---|---|
Arena Elo Human preference ranking via blind comparisons | 1480 | #15 / 51 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 87.4% | #15 / 51 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 45.5% | #22 / 58 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 87.3% | #30 / 61 |
HLEArtificial Analysis Challenging multidisciplinary questions evaluated by Artificial Analysis | 40.7% | #40 / 100 |
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 88.4% | #50 / 96 |
| Benchmark / source | Score |
|---|---|
HLE Artificial AnalysisMuse Spark | 40.7% |
MMMU-Pro Artificial AnalysisMuse Spark | 80.5% |
Terminal-Bench 2.1 Artificial AnalysisMuse Spark | 62.2% |
AA-LCR Artificial AnalysisMuse Spark | 78% |
CritPt Artificial AnalysisMuse Spark | 11.3% |
IFBench Artificial AnalysisMuse Spark | 75.9% |