DeepSeek-OCR.aiConvert PDF and Image to clean text and tables with DeepSeek-OCR
Extract text, tables, and layouts from scans, photos, and PDFs. Get accurate results in Markdown, CSV, or plain text — with cutting‑edge AI DeepSeek‑OCR.
Features
Everything you need to turn images and PDFs into usable text.
Processes pages quickly and keeps formatting stable for consistent results.
Handles multi‑page PDFs and large scans without skipping content.
Efficient processing uses fewer tokens so you pay less.
Extracts clean CSV tables and preserves page structure where it matters.
Understands columns, headers, captions, and figures for clean output.
Recognizes English, Chinese, and mixed documents with proper punctuation.
How to use
Quick starters for common OCR tasks and grounding.
# document: <image>
<|grounding|>Convert the document to markdown.# other image: <image>
<|grounding|>OCR this image.# without layouts: <image>
Free OCR.# figures in document: <image>
Parse the figure.# general: <image>
Describe this image in detail.# rec: <image>
Locate <|ref|>xxxx<|/ref|> in the image.Visualizations
Sample OCR visualizations and outputs from DeepSeek‑OCR.
Real‑World Example
A practical document image demonstrating DeepSeek‑OCR on real input.
Pricing
Simple plans for online and API access.
10 uses/day (online + API)
- • Image/PDF OCR
- • Output formats: Text, Markdown, CSV
- • Max file size: 10MB
- • Rate‑limited queue
- • Community support
1000 uses (online + API)
- • Priority processing
- • Larger files up to 25MB
- • Batch uploads
- • Email support
- • Export formats: Text/Markdown/CSV
Unlimited usage, SSO, SLA
- • Dedicated support & onboarding
- • On‑prem or private cloud
- • SSO/SAML, audit logs
- • SOC 2 / GDPR alignment
- • Custom models & regional routing