RUN THIS LLM
Search local LLM hardware requirements
DeepSeek VL2 Small
DeepSeek · 16.1B MoE · Vision
16.1B MoE vision-language model with 2.8B active params. Excellent document understanding, chart QA, and grounding for its activated size.
VRAM Requirements
| Quantization | VRAM |
|---|---|
| Q4_K_M (smallest) | 9.7 GB |
| Q8_0 (balanced) | 17.7 GB |
| FP16 (full quality) | 32.2 GB |
Specifications
- Parameters: 16.1B MoE (2.8B active per token)
- Category: Vision
- Max context: 4K tokens
- System RAM: 16 GB minimum
- HuggingFace: deepseek-ai/deepseek-vl2-small
Loading interactive analysis...