Skip to main content
Ollama ExplorerBeta
VisionbeginnerVision

deepseek-ocr

DeepSeekDeepSeek

DeepSeek-OCR is a vision-language model that can perform token-efficient OCR.

206K pullsUpdated Nov 26, 20253 tags8K context

Quick start

ollama run deepseek-ocr

Available sizes

TagSizeQuantizationContextMin RAM
deepseek-ocr:latest6.7GBq4_k_m8K context8.4 GB

Strengths & Limitations

Strengths

  • Token-efficient OCR
  • Vision-language capabilities
  • Performs Optical Character Recognition

Related models