OCR

Extract text, tables, and form fields from documents and images using AI-powered OCR.

POST /v1/ocr

Overview

The OCR API converts scanned documents and images into structured text. Powered by Mistral's document AI, it supports:

• Text extraction - Accurate text from documents and images
• Table extraction - Structured table data in JSON
• Form field extraction - Key-value pairs from forms
• Multi-page support - Process entire PDFs
• Multilingual - 100+ languages

Request

cURL

curl -X POST https://api.gateflow.ai/v1/ocr \
  -H "Authorization: Bearer gw_prod_..." \
  -F "file=@document.pdf" \
  -F "extract_tables=true" \
  -F "extract_forms=true"

Python

import requests

response = requests.post(
    "https://api.gateflow.ai/v1/ocr",
    headers={"Authorization": "Bearer gw_prod_..."},
    files={"file": open("document.pdf", "rb")},
    data={
        "extract_tables": True,
        "extract_forms": True,
        "pages": "1-5"
    }
)

TypeScript

const formData = new FormData();
formData.append('file', fileBlob, 'document.pdf');
formData.append('extract_tables', 'true');
formData.append('extract_forms', 'true');

const response = await fetch('https://api.gateflow.ai/v1/ocr', {
  method: 'POST',
  headers: {
    'Authorization': 'Bearer gw_prod_...',
  },
  body: formData,
});

Parameters

Parameter	Type	Required	Default	Description
file	file	Yes	-	Document file (PDF, PNG, JPG, TIFF)
extract_tables	boolean	No	true	Extract tables as structured data
extract_forms	boolean	No	true	Extract form field key-value pairs
pages	string	No	all	Page range (e.g., 1-5, all)
language_hints	array	No	null	Language codes to improve accuracy
output_format	string	No	json	Output: json, text, markdown

Response

{
  "document_id": "550e8400-e29b-41d4-a716-446655440000",
  "total_pages": 3,
  "pages": [
    {
      "page_number": 1,
      "text": "INVOICE\n\nInvoice Number: INV-2024-001\nDate: January 15, 2024\n...",
      "tables": [
        {
          "table_id": "table_1",
          "rows": [
            ["Item", "Quantity", "Price"],
            ["Widget A", "10", "$50.00"],
            ["Widget B", "5", "$25.00"]
          ],
          "bounding_box": {"x": 50, "y": 200, "width": 400, "height": 150}
        }
      ],
      "forms": [
        {"key": "Invoice Number", "value": "INV-2024-001", "confidence": 0.98},
        {"key": "Date", "value": "January 15, 2024", "confidence": 0.95}
      ],
      "confidence": 0.96
    }
  ],
  "processing_time_ms": 1250,
  "_gateway": {
    "provider": "mistral",
    "model": "mistral-ocr-latest",
    "cost": {"usd": 0.03}
  }
}

Response Fields

Field	Type	Description
document_id	string	Unique document identifier
total_pages	integer	Number of pages processed
pages	array	Per-page extraction results
pages[].text	string	Extracted plain text
pages[].tables	array	Extracted table data
pages[].forms	array	Extracted form fields
pages[].confidence	number	Overall confidence score (0-1)
processing_time_ms	integer	Processing time
_gateway	object	GateFlow metadata

Supported Formats

Format	Extensions	Max Size	Notes
PDF	.pdf	50MB	Multi-page supported
PNG	.png	20MB	Single image
JPEG	.jpg, .jpeg	20MB	Single image
TIFF	.tiff, .tif	50MB	Multi-page supported
BMP	.bmp	20MB	Single image
WebP	.webp	20MB	Single image

Language Support

GateFlow OCR supports 100+ languages. Use language_hints to improve accuracy:

{
  "file": "<binary>",
  "language_hints": ["en", "nl", "de"]
}

Common language codes:

• en - English
• es - Spanish
• fr - French
• de - German
• nl - Dutch
• zh - Chinese
• ja - Japanese
• ko - Korean
• ar - Arabic

Use Cases

Invoice Processing

response = requests.post(
    "https://api.gateflow.ai/v1/ocr",
    headers={"Authorization": "Bearer gw_prod_..."},
    files={"file": open("invoice.pdf", "rb")},
    data={
        "extract_tables": True,
        "extract_forms": True
    }
)

# Extract invoice fields
for page in response.json()["pages"]:
    for field in page["forms"]:
        if field["key"].lower() in ["invoice number", "total", "date"]:
            print(f"{field['key']}: {field['value']}")

Medical Record Processing

# Process medical record with PII awareness
response = requests.post(
    "https://api.gateflow.ai/v1/ocr",
    headers={"Authorization": "Bearer gw_prod_..."},
    files={"file": open("medical_record.pdf", "rb")},
    data={
        "extract_forms": True,
        "language_hints": ["en"]
    }
)

# Automatically detects and flags PHI fields

Batch Processing

import asyncio
import aiohttp

async def process_documents(files):
    async with aiohttp.ClientSession() as session:
        tasks = []
        for file_path in files:
            with open(file_path, "rb") as f:
                data = aiohttp.FormData()
                data.add_field('file', f.read(), filename=file_path)
                data.add_field('extract_tables', 'true')

                task = session.post(
                    "https://api.gateflow.ai/v1/ocr",
                    headers={"Authorization": "Bearer gw_prod_..."},
                    data=data
                )
                tasks.append(task)

        results = await asyncio.gather(*tasks)
        return [await r.json() for r in results]

Error Codes

Code	Description
400	Invalid file format or corrupted file
401	Invalid API key
413	File too large
422	Could not process document
429	Rate limit exceeded
500	OCR service error

Pricing

Operation	Cost
Per page	$0.01
Tables	Included
Forms	Included

See Also

• Documents API - Store OCR results for RAG
• Search API - Search extracted text
• PII Detection - Handle sensitive data