DeepSeek VL2 Small
Multimodal
Zero-eval
#2MMBench-V1.1
#2MME
#3MMT-Bench
by DeepSeek
+
+
+
+
About
DeepSeek-VL2-Small was created as a compact vision-language variant, designed to bring multimodal capabilities to applications with limited computational resources. Built to provide visual and textual understanding in a more efficient package, it serves use cases requiring practical deployment of vision-language AI.
+
+
+
+
Timeline
AnnouncedDec 13, 2024
ReleasedDec 13, 2024
+
+
+
+
Specifications
Capabilities
Multimodal
+
+
+
+
License & Family
License
deepseek
Performance Overview
Performance metrics and category breakdown
Overall Performance
14 benchmarks
Average Score
69.6%
Best Score
92.3%
High Performers (80%+)
6+
+
+
+
All Benchmark Results for DeepSeek VL2 Small
Complete list of benchmark scores with detailed information
| DocVQA | multimodal | 0.92 | 92.3% | Self-reported | |
| ChartQA | multimodal | 0.84 | 84.5% | Self-reported | |
| TextVQA | multimodal | 0.83 | 83.4% | Self-reported | |
| OCRBench | multimodal | 0.83 | 83.4% | Self-reported | |
| MMBench | multimodal | 0.80 | 80.3% | Self-reported | |
| AI2D | multimodal | 0.80 | 80.0% | Self-reported | |
| MMBench-V1.1 | multimodal | 0.79 | 79.3% | Self-reported | |
| InfoVQA | multimodal | 0.76 | 75.8% | Self-reported | |
| RealWorldQA | multimodal | 0.65 | 65.4% | Self-reported | |
| MMT-Bench | multimodal | 0.63 | 62.9% | Self-reported |
Showing 1 to 10 of 14 benchmarks
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+