DeepSeek-OCR Demo

Upload an image to extract text using DeepSeek-OCR model. Supports various document types and handwriting recognition.

Model Sizes:

  • Tiny: Fastest, lower accuracy (512x512)
  • Small: Fast, good accuracy (640x640)
  • Base: Balanced performance (1024x1024)
  • Large: Best accuracy, slower (1280x1280)
  • Gundam (Recommended): Optimized for documents (1024 base, 640 image, crop mode)
Model Size
Task Type

Returns only plain text, but might be faster. Uncheck to get annotated image and markdown.
Examples
Upload Image Model Size Task Type