VideoLLaMA2-72B: benchmark results
Provider: Alibaba. Released 2024-08-14.
Aggregate Rating 1575 ± 33, rank #1323 of 10916 rated models, from 22 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MLVU-Test - Action Order | 42.9 | Multiple-choice accuracy (%) | 85.3 |
| MLVU-Test - Ego Reasoning | 54.7 | Multiple-choice accuracy (%) | 83.8 |
| MLVU-Test - Generation | 3.95 | Generation score, GPT-4 judged (0-10) | 78.1 |
| MLVU-Test - Anomaly Recognition | 53.8 | Multiple-choice accuracy (%) | 75 |
| MLVU-Test | 45.6 | Multiple-choice accuracy (%) | 73.5 |
| MLVU-Test - Plot QA | 54 | Multiple-choice accuracy (%) | 73.5 |
| Vinoground (Text Score) | 36.2 | Text Score | 69.6 |
| MLVU-Dev | 61.2 | Multiple-choice accuracy (%) | 68.4 |
| MLVU-Test - Topic Reasoning | 80.2 | Multiple-choice accuracy (%) | 66.2 |
| AffectSim (Active Observation) | 14.08 | Macro-F1 (%; A-Obs: the view reached by the authors' two-sta | 65.2 |
| MLVU-Test - Tutorial QA | 32.6 | Multiple-choice accuracy (%) | 63.2 |
| MLVU-Test - Sports QA | 38.9 | Multiple-choice accuracy (%) | 61.8 |
Interactive version: theaggregate.ai/model?slug=videollama2-72b · How It Works · Data refreshed daily, snapshot 2026-11-08.