Skip to content

Benchmark

BLINK - Multi-view_Reasoning

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

49.8%
Mean
53.4%
Median
6.3%
Std dev
41.4%
10th percentile
57.9%
90th percentile
41.4%
Min

Leaderboard

#ModelScore
1Claude 3 OpusMulti-view_Reasoning· EvalAI data57.9%
2Qwen-VL MaxMulti-view_Reasoning· EvalAI data53.4%
3LLaVA 1.6 34BMulti-view_Reasoning· EvalAI data46.6%
4Gemini ProMulti-view_Reasoning· EvalAI data41.4%