Llava-OneVision-Qwen2-7B-SI: benchmark results
Provider: LLaVA Team. Access: Open.
Aggregate Rating 1462 ± 52, rank #3345 of 10516 rated models, from 11 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MIOH - Existence | 47.9 | Accuracy (%; existence questions about whether objects are p | 51.9 |
| M3-AD-Bench - Electronic | 55.9 | Balanced accuracy (%) of binary anomaly detection (is the im | 50 |
| MIOH - Position | 34.4 | Accuracy (%; position questions about object positions acros | 50 |
| M3-AD-Bench - Workpiece | 57.7 | Balanced accuracy (%) of binary anomaly detection (is the im | 43.8 |
| MIOH | 32.5 | Accuracy (%; mean of the existence, counting, attribute and | 40.7 |
| M3-AD-Bench | 60.4 | Balanced accuracy (%) of binary anomaly detection (is the im | 37.5 |
| MIOH - Counting | 23.2 | Accuracy (%; counting questions about how many objects there | 33.3 |
| M3-AD-Bench - Anomaly Type F1 | 24.2 | Anomaly-type hard micro-F1 (0-1, times 100): a predicted ano | 31.2 |
| M3-AD-Bench - Texture | 67.6 | Balanced accuracy (%) of binary anomaly detection (is the im | 27.8 |
| MIOH - Attribute | 24.5 | Accuracy (%; attribute questions about object attributes acr | 14.8 |
| SpaceDG | 33.8 | Avg. (self-reported) | 10 |
Interactive version: theaggregate.ai/model?slug=llava-onevision-qwen2-7b-si · How It Works · Data refreshed daily, snapshot 2026-11-21.