ShareGPT4Video-8B: benchmark results
Provider: LLaVA Team.
Aggregate Rating 1454 ± 30, rank #3578 of 10916 rated models, from 62 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MLVU-Test - Ego Reasoning | 45.3 | Multiple-choice accuracy (%) | 67.6 |
| MLVU-Test - Sports QA | 38.9 | Multiple-choice accuracy (%) | 61.8 |
| MLVU-Test - Tutorial QA | 25.6 | Multiple-choice accuracy (%) | 51.5 |
| MLVU-Test - Needle QA | 31.7 | Multiple-choice accuracy (%) | 50 |
| MLVU-Test - Plot QA | 38 | Multiple-choice accuracy (%) | 48.5 |
| MLVU-Test | 33.8 | Multiple-choice accuracy (%) | 47.1 |
| INFACT (Factuality) - Clean Accuracy | 48.9 | Accuracy (0-1, shown times 100) on clean video (Mode I) for | 46.2 |
| EgoMonth - Event Time | 45.3 | Accuracy (%; 117 event time questions; four-option multiple- | 45.5 |
| MLVU-Dev | 46.4 | Multiple-choice accuracy (%) | 44.7 |
| MLVU-Test - Topic Reasoning | 73.6 | Multiple-choice accuracy (%) | 44.1 |
| MLVU-Test - Generation | 3.63 | Generation score, GPT-4 judged (0-10) | 43.8 |
| EgoMonth - Direction Judgement | 41.7 | Accuracy (%; 36 direction judgement questions; four-option m | 40.9 |
Interactive version: theaggregate.ai/model?slug=sharegpt4video-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.