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

BenchmarkScoreMetricPercentile
MIOH - Existence47.9Accuracy (%; existence questions about whether objects are p51.9
M3-AD-Bench - Electronic55.9Balanced accuracy (%) of binary anomaly detection (is the im50
MIOH - Position34.4Accuracy (%; position questions about object positions acros50
M3-AD-Bench - Workpiece57.7Balanced accuracy (%) of binary anomaly detection (is the im43.8
MIOH32.5Accuracy (%; mean of the existence, counting, attribute and 40.7
M3-AD-Bench60.4Balanced accuracy (%) of binary anomaly detection (is the im37.5
MIOH - Counting23.2Accuracy (%; counting questions about how many objects there33.3
M3-AD-Bench - Anomaly Type F124.2Anomaly-type hard micro-F1 (0-1, times 100): a predicted ano31.2
M3-AD-Bench - Texture67.6Balanced accuracy (%) of binary anomaly detection (is the im27.8
MIOH - Attribute24.5Accuracy (%; attribute questions about object attributes acr14.8
SpaceDG33.8Avg. (self-reported)10

Interactive version: theaggregate.ai/model?slug=llava-onevision-qwen2-7b-si · How It Works · Data refreshed daily, snapshot 2026-11-21.