Meta's first proprietary model from Superintelligence Labs, non-open-weight
| Benchmark | Score | Rank |
|---|---|---|
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 87.4% | #6 / 39 |
Arena Elo Human preference ranking via blind comparisons | 1480 | #16 / 52 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 87.3% | #16 / 38 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 45.5% | #16 / 48 |
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 88.4% | #19 / 64 |