Google Lens OCR API: Sub-second Image to Text

Extract text from any image with exact word-level bounding boxes and absolute pixel coordinates. Powered by the official Google Lens engine for sub-second, multi-language OCR under 500ms. No browser required.

Get fully structured, line-by-line paragraph, line, and word extractions with absolute pixel width, height, and coordinates, fully optimized for price intelligence, document indexing, and AI RAG pipelines.

---

What can you do with it?

• I am an AI developer building a RAG pipeline and looking to quickly extract high-fidelity text segments, tables, and paragraphs from scanned documents or PDFs in under 500ms.
• I am an automation engineer looking to programmatically OCR receipts, invoices, and menus with precise pixel bounding-box coordinates for automatic layout reconstructions.
• I am a translation vendor looking to hint specific languages and instantly translate image text blocks to another target locale.

---

How to extract text from images step-by-step

1. Create a free Apify account.
2. Open the Google Lens OCR API in the Apify Store.
3. Enter your target Image URL (e.g. any public JPEG, PNG, or WebP link).
4. Configure optional filters like Language Hint or Translate To.
5. Click Start and download your structured OCR text dataset as JSON, CSV, or Excel.

API quick start

Trigger runs programmatically with a single POST request in under 500ms:

curl --request POST \
  --url "https://google-lens-ocr.apify.actor/ocr?token=YOUR_API_TOKEN" \
  --header 'Content-Type: application/json' \
  --data '{
    "imageUrl": "https://tesseract.projectnaptha.com/img/eng_bw.png",
    "outputDetail": "full"
  }'

---

Input

Field	Type	Required	Description
imageUrl	string	No	The public URL of a single image to extract text from. Either imageUrl or imageUrls must be provided.
imageUrls	array of URLs	No	Paste multiple image URLs to extract in a single bulk-batch run.
outputDetail	enum	No	Level of structural detail to output: full (lines + boxes), paragraphs, lines, words, text_only.
language	string	No	ISO 639-1 language code hint to improve OCR accuracy (e.g. en, de, fr, ja, ar).
region	enum	No	Select country region hint to improve address, phone number, and date parser formats.
translateTo	string	No	Optional ISO 639-1 language code to automatically translate detected text block (e.g. es, de).

---

Output

Each run produces one result with these fields. Bounding boxes include absolute pixel coordinates (x, y, width, height) for immediate use in image annotations.

{
  "imageUrl": "https://tesseract.projectnaptha.com/img/eng_bw.png",
  "language": "en",
  "fullText": "Mild Splendour of the various-vested Night!\nMother of wildly-working visions! hail!",
  "lines": [
    {
      "text": "Mild Splendour of the various-vested Night!",
      "x": 30,
      "y": 33,
      "width": 1265,
      "height": 77
    },
    {
      "text": "Mother of wildly-working visions! hail!",
      "x": 99,
      "y": 109,
      "width": 1099,
      "height": 76
    }
  ],
  "scrapedAt": "2026-06-15T11:14:58.726Z"
}

Field reference

Field	Type	Description
imageUrl	string	Public URL of the parsed image.
language	string	Detected language code.
fullText	string	Complete extracted text block joined by newlines.
lines	array	Lines of text with absolute pixel coordinates (x, y, width, height).
paragraphs	array	Paragraph elements with absolute pixel coordinates.
scrapedAt	string	ISO timestamp of when the image was processed.

---

Pricing and cost

Pricing is pay-per-result and is billed per successfully processed image. Failed runs cost absolutely nothing.

• Rate: $4.99 per 1,000 images ($0.00499 per image).
• 100 images typically cost $0.49.
• 1,000 images cost exactly $4.99.
• 10,000 images cost exactly $49.90.
• No fixed monthly subscriptions or hidden maintenance fees.

---

Tips and best practices

• Avoid browser cold starts. By deploying in Standby mode, this Actor remains alive in the background. You receive sub-second OCR responses (typically under 200-500ms) with zero initialization delay!
• Keep dimensions reasonable. Images are automatically resized to under 1200x1200px before uploading. This speeds up transfer latency and avoids large binary payload blocks while preserving 100% OCR accuracy.

---

FAQ

Does it require a browser to execute?

No. Unlike browser-rendered scrapers, this Actor connects directly to Google's native binary API. This guarantees sub-second execution speeds, zero browser memory overhead, and extremely low running costs.

Are all image formats supported?

Yes. The Actor supports JPEG, PNG, WebP, BMP, TIFF, and HEIC formats.

Is handwriting recognized?

Yes. The Actor uses the same advanced neural OCR engine as Google Lens in Chromium. Accuracy is exceptionally high for both printed text and clear handwriting.

---

Target SEO keywords

• Google Lens OCR API
• image to text API
• extract text from image
• word bounding box coordinates
• Chromium OCR unblocked

---

Disclaimers and support

• Disclaimer: This Actor connects to publicly available image uploading services. Users are responsible for ensuring they have the rights to process the images they submit.
• Support: Open an issue from the Actor page in the Apify Console for bug reports or feature requests.