OCR that actually works
Extract text from any document — receipts, invoices, passports, handwritten notes — with 99.2% accuracy and a simple REST API. No preprocessing needed.
- 99.2% accuracy on printed text
- ~180ms median latency
- 100+ languages
- Layout-aware extraction
Live demo — powered by Claude
Upload any document and see real extraction results in seconds.
Input
Drop a document or image
or click to browse · JPEG, PNG, WebP · max 5MB
Extracted text
Upload an image to see extracted text
Built for production
Sub-200ms latency
Median latency under 180ms globally with edge deployment across 12 regions.
100+ languages
Latin, Arabic, CJK, Devanagari, and more. Automatic language detection.
Noise & skew tolerant
Handles rotated documents, low resolution, watermarks, and complex backgrounds.
Layout preservation
Maintains column structure, tables, and spatial relationships in output.
One API call. That’s it.
No training, no model configuration, no infrastructure to manage. POST your document, receive structured text. Our SDK handles retries, rate limits, and error handling.
curl -X POST https://api.quantilence.com/v1/ocr \
-H "Authorization: Bearer qk_live_..." \
-F "file=@document.jpg"{
"text": "Invoice #INV-2024-001\nDate: 2024-01-15\nBill To: Acme Corp...",
"confidence": 0.982,
"latencyMs": 178,
"wordCount": 247,
"detectedLanguage": "en",
"pages": 1
}