📄 OCR — Text from Images

Extract all visible text from any image using Claude's vision capabilities. Handles printed text, handwriting, receipts, invoices, screenshots, and documents. Supports JPEG, PNG, GIF, and WebP. Input can be an image URL or base64-encoded data. Returns extracted text, detected language, and confidence score.

Request Body

Parameter	Type	Required	Description	Default
`image`	string	required	Public image URL (https://) OR base64-encoded image (with or without data URI prefix)	—
`language`	string	optional	ISO 639-1 language hint (e.g. en, fr, ja) or 'auto' to detect	`auto`

Response Fields

Field	Type	Description
`text`	string	All text extracted from the image (empty string if none found)
`language`	string	Detected or confirmed ISO 639-1 language code
`confidence`	number	Extraction confidence score 0.0–1.0
`source`	string	Input type used: url \| base64
`processedAt`	string	ISO 8601 timestamp of when OCR was performed

Code Examples

cURL

bash

# From URL
curl -X POST https://sparkforge.sh/api/ocr \
  -H "Content-Type: application/json" \
  -H "X-PAYMENT: <x402-payment-token>" \
  -d '{
    "image": "https://example.com/invoice.jpg",
    "language": "en"
  }'

# From base64
curl -X POST https://sparkforge.sh/api/ocr \
  -H "Content-Type: application/json" \
  -H "X-PAYMENT: <x402-payment-token>" \
  -d '{"image": "data:image/png;base64,iVBORw0KGgo..."}'

Python

python

import requests, base64

# From URL
response = requests.post(
    "https://sparkforge.sh/api/ocr",
    headers={"Content-Type": "application/json", "X-PAYMENT": "<token>"},
    json={"image": "https://example.com/receipt.jpg", "language": "en"}
)

# From file
with open("receipt.jpg", "rb") as f:
    b64 = base64.b64encode(f.read()).decode()

response = requests.post(
    "https://sparkforge.sh/api/ocr",
    headers={"Content-Type": "application/json", "X-PAYMENT": "<token>"},
    json={"image": b64}
)

data = response.json()
print(data["text"])
print(f"Language: {data['language']} | Confidence: {data['confidence']:.0%}")

TypeScript / Node.js

typescript

import * as fs from "fs";

// From URL
const res = await fetch402("https://sparkforge.sh/api/ocr", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({ image: "https://example.com/receipt.jpg" }),
});

const { text, language, confidence } = await res.json();
console.log(`Extracted (${language}, ${(confidence * 100).toFixed(0)}% confidence):`, text);

Example Response

json

{
  "text": "Invoice #4821\nDate: 2024-03-01\nBill To: Acme Corp\n\nItem           Qty  Price\nWidget Pro      2    $25.00\nShipping               $5.00\n\nTotal Due: $55.00",
  "language": "en",
  "confidence": 0.97,
  "source": "url",
  "processedAt": "2026-02-28T16:00:00.000Z"
}

💳 Payment via x402

This endpoint costs $0.02 per call, paid in USDC on Base, Polygon, or Solana. Use the @x402/fetch library to handle payments automatically, or implement the x402 handshake manually.

bash

# Free demo mode (no payment)
curl https://sparkforge.sh/api/ocr?demo=true

# Or try via MCP (Claude Desktop, Cursor, Windsurf)
# Add to mcp config: { "url": "https://sparkforge.sh/api/mcp" }

←

File Format Converter

Site Crawler

→