Cosmos3-Super: benchmark results
Provider: Other.
Aggregate Rating 1776 ± 46, rank #132 of 10916 rated models, from 18 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| TAR-Bench | 48.8 | Mean score (%; unweighted mean of the ten task scores: accur | 100 |
| TAR-Bench - Binary QA | 89.8 | Accuracy (%; yes/no questions about the anomaly, one positiv | 100 |
| TAR-Bench - Binary QA with Explanation | 60 | Scaled BERTScore F1 (%; a yes/no answer with an explanation, | 100 |
| TAR-Bench - Causal Linkage | 31 | Scaled BERTScore F1 (%; an explanation of what caused the an | 100 |
| TAR-Bench - Multiple-Choice QA | 88.8 | Accuracy (%; multiple-choice questions about the anomaly; ze | 100 |
| TAR-Bench - Open-Ended QA | 40.4 | Scaled BERTScore F1 (%; free-form answers compared with the | 100 |
| TAR-Bench - Scene Description | 28.8 | Scaled BERTScore F1 (%; a physical description of the scene, | 100 |
| TAR-Bench - Temporal Description | 30.3 | Scaled BERTScore F1 (%; a description of what happens in a g | 100 |
| TAR-Bench - Video Summarization | 22.4 | Scaled BERTScore F1 (%; a summary of the events in the clip, | 100 |
| TAR-Bench - Multiple-Choice QA with Explanation | 71.2 | Scaled BERTScore F1 (%; a multiple-choice answer with an exp | 90 |
| VANTAGE-Bench | 63.62 | Overall (%) | 82.4 |
| VANTAGE-Bench - Spatial | 78.71 | Spatial (%) | 82.4 |
Interactive version: theaggregate.ai/model?slug=cosmos3-super · How It Works · Data refreshed daily, snapshot 2026-11-08.