LLaVA-Next-8B: benchmark results
Provider: LLaVA Team. Released 2024-05-10. Access: Open.
Aggregate Rating 1522 ± 39, rank #2046 of 10916 rated models, from 53 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Cambrian-S Release Image - GQA | 65.2 | Accuracy (%) | 100 |
| Cambrian-S Release Image - MME Perception | 1603.7 | MME perception score | 83.3 |
| MUSE - Object Classification | 38 | Accuracy (%) | 75.9 |
| HM-Bench (PCA) | 39.03 | Average accuracy (%) over all tasks of HM-Bench's 19,337 hyp | 60 |
| MUSE - Activity Localization | 44.5 | Accuracy (%) | 51.7 |
| Cambrian-S Release Image - AI2D | 71.6 | Accuracy (%) | 50 |
| Cambrian-S Release Image - ChartQA | 69.5 | Accuracy (%) | 50 |
| Cambrian-S Release Image - General Average | 72.5 | Accuracy (%) | 50 |
| SpatialThinker Release - 3DSRMultiObject | 43.4 | Accuracy (%) | 47.6 |
| SpatialThinker Release - 3DSROrientation | 36.1 | Accuracy (%) | 47.6 |
| HM-Bench (Report) | 36.71 | Average accuracy (%) over all tasks of HM-Bench's 19,337 hyp | 46.7 |
| SpatialThinker Release - 3DSRBench | 48.4 | Accuracy (%) | 44.4 |
Interactive version: theaggregate.ai/model?slug=llava-next-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.