Skip to content

Benchmark

BLINK - Functional_Correspondence

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

28.5%
Mean
30.8%
Median
3.8%
Std dev
22.3%
10th percentile
32.3%
90th percentile
22.3%
Min

Leaderboard

#ModelScore
1Gemini ProFunctional_Correspondence· EvalAI data32.3%
2LLaVA 1.6 34BFunctional_Correspondence· EvalAI data30.8%
3Qwen-VL MaxFunctional_Correspondence· EvalAI data28.5%
4Claude 3 OpusFunctional_Correspondence· EvalAI data22.3%