Skip to content

Benchmark

BLINK - Spatial_Relation

Vision benchmarks measure how well models understand images. Scores are percentages.

Models measured
4
Scores recorded
4
Sources
1
IQ weight
No weight published

Score distribution

69.6%
Mean
76.2%
Median
8.1%
Std dev
57.3%
10th percentile
77.6%
90th percentile
57.3%
Min
57.3%77.6%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Qwen-VL MaxSpatial_Relation· EvalAI data77.6%
2LLaVA 1.6 34BSpatial_Relation· EvalAI data76.2%
3Gemini ProSpatial_Relation· EvalAI data67.1%
4Claude 3 OpusSpatial_Relation· EvalAI data57.3%