mPLUG-Owl3-7B: benchmark results
Provider: Alibaba.
Aggregate Rating 1476 ± 22, rank #2716 of 10916 rated models, from 49 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CLIP-CC-Bench | 66 | Mean HM-CF similarity (0-100): harmonic mean of paragraph-le | 93.8 |
| MVPBench - Video Brightness Evaluation | 29.29 | MVPBench normalized proficiency score (accuracy minus chance | 92.9 |
| LongVideoBench | 60.1 | Test Total Accuracy (%) | 73.5 |
| MVPBench - Dance Style Matching | 9.28 | MVPBench normalized proficiency score (accuracy minus chance | 71.4 |
| PhysBench | 42.8 | Overall Accuracy (%) | 62.5 |
| FineBench - Object Manipulation | 75.2 | Accuracy (%; four-option multiple-choice questions on how a | 60 |
| MVPBench - Caption-Free Temporal Splicing | -4.46 | MVPBench normalized proficiency score (accuracy minus chance | 57.1 |
| MVPBench - Caption-Guided Temporal Splicing | 11.5 | MVPBench normalized proficiency score (accuracy minus chance | 57.1 |
| LifeEval (Multiple-Choice) - Contextual Knowledge Retrieval | 66.41 | Accuracy (%) on the 390 multiple-choice Contextual Knowledge | 52 |
| CoSimUE | 44.73 | Pair accuracy (%; both the original and the edited image ans | 46.7 |
| CoSimUE - Co-occurrence | 44.67 | Pair accuracy (%; 338 co-occurrence question pairs) | 46.7 |
| CoSimUE - Similarity | 44.83 | Pair accuracy (%; 174 similarity question pairs) | 46.7 |
Interactive version: theaggregate.ai/model?slug=mplug-owl3-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.