RUN THIS LLM
Search local LLM hardware requirements
DeepSeek VL2 Tiny
DeepSeek · 3.4B MoE · Vision
Compact 3.4B MoE vision-language model with 1B active params. Strong OCR, document QA, and visual grounding at edge-device size.
VRAM Requirements
| Quantization | VRAM |
|---|---|
| Q4_K_M (smallest) | 2 GB |
| Q8_0 (balanced) | 3.7 GB |
| FP16 (full quality) | 6.8 GB |
Specifications
- Parameters: 3.4B MoE (1B active per token)
- Category: Vision
- Max context: 4K tokens
- System RAM: 4 GB minimum
- HuggingFace: deepseek-ai/deepseek-vl2-tiny
Loading interactive analysis...