llava-onevision-qwen2-72B-ov-sft: benchmark results

Provider: LLaVA Team. Access: Open.

Aggregate Rating 1522 ± 46, rank #1957 of 10556 rated models, from 14 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
ColorBench51.9P&R Overall accuracy (%)96.7
KOFFVQA - Hallucination and Robustness86Score (%)94.4
ColorBench - Color Robustness80.3Color Robustness (%)90
KOFFVQA - Relationship71Score (%)82.7
KOFFVQA - Object Attributes77.83Score (%)71.6
KOFFVQA - Commonsense Reasoning79.33Score (%)62.3
KOFFVQA - Recognition65Score (%)45.1
KOFFVQA - Graph and Chart Understanding54Score (%)40.7
KOFFVQA54.15Overall Score (%)38.3
KOFFVQA - Korean Recognition5Score (%)22.8
KOFFVQA - Table Understanding18.33Score (%)18.5
KOFFVQA - Document Understanding23Score (%)15.4

Interactive version: theaggregate.ai/model?slug=llava-onevision-qwen2-72b-ov-sft · How It Works · Data refreshed daily, snapshot 2026-11-17.