Long-LLaVA: benchmark results
Provider: Other.
Aggregate Rating 1482 ± 50, rank #2646 of 10916 rated models, from 14 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SceneBench - Comment Prediction | 79.7 | Accuracy (%) on SceneBench's 297 comment prediction items (c | 75 |
| SceneBench - Clip QA | 57.1 | Accuracy (%) on SceneBench's 368 clip qa items (questions ab | 61.1 |
| SceneBench - Scene QA | 26.1 | Accuracy (%) on SceneBench's 3,311 scene qa items (questions | 47.2 |
| SceneBench | 52.9 | Average accuracy (%) over SceneBench's six tasks (title and | 44.4 |
| SceneBench - Title Prediction | 96.2 | Accuracy (%) on SceneBench's 345 title prediction items (cho | 27.8 |
| Video-MME - Long Videos (w/ subs) | 50.3 | Accuracy (%) | 26 |
| Video-MME - Medium Videos (w/o subs) | 51.4 | Accuracy (%) | 24 |
| Video-MME - Short Videos (w/ subs) | 66.2 | Accuracy (%) | 24 |
| SceneBench - Inverse VQA | 33 | Accuracy (%) on SceneBench's 485 inverse vqa items (choosing | 22.2 |
| Video-MME - Medium Videos (w/ subs) | 54.7 | Accuracy (%) | 22 |
| Video-MME - Short Videos (w/o subs) | 61.9 | Accuracy (%) | 22 |
| VideoMME w sub. | 57.1 | Score (imported) | 20 |
Interactive version: theaggregate.ai/model?slug=long-llava · How It Works · Data refreshed daily, snapshot 2026-11-08.