Benchmark
BLINK - Art_Style
Vision benchmarks measure how well models understand images. Scores are percentages.
All vision benchmarksScores from EvalAI data.
- Models measured
- 4
- Scores recorded
- 4
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 48.7%
- Mean
- 49.6%
- Median
- 8.2%
- Std dev
- 37.6%
- 10th percentile
- 60.7%
- 90th percentile
- 37.6%
- Min
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Claude 3 OpusArt_Style· EvalAI data | 60.7% |
| 2 | Gemini ProArt_Style· EvalAI data | 49.6% |
| 3 | LLaVA 1.6 34BArt_Style· EvalAI data | 47% |
| 4 | Qwen-VL MaxArt_Style· EvalAI data | 37.6% |