Google Images Scraper — Full-Size Image URLs & Sources

Search Google Images and get back the full-size image URLs, not just thumbnails. No API key, no Google Cloud project, no headless browser.

What you get

Per image:

Field	Coverage	Description
image_url	Always	The full-size image, hosted at its original source
width / height	Always	Original dimensions in pixels
title	Most results	The image's title as Google shows it
source_url	Most results	The page the image appears on
thumbnail_url	Rarely	Google's thumbnail, when it serves one as a URL

On partial fields. Google returns image search results as a page that doesn't pair every image with its title in a machine-readable way. This Actor matches them by the image's host domain, which resolves most results — around 7 in 10 on a typical query. Where an image is served from a CDN unrelated to its source (a history.com article whose photo sits on cloudinary.com, say), title and source_url are left null rather than guessed. A blank field is recoverable; a confidently wrong one silently corrupts your dataset.

Every result always carries image_url and dimensions, so if that is what you need, coverage is complete.

Results land in the run's dataset and export to JSON, CSV, Excel, XML or via the API.

Pricing

You pay for	Price
Each image scraped	$0.12 per 1,000
Each results page requested	$0.004
Actor start	$0.05 per 1,000 runs

Google returns about 100 images per page, so 1,000 images works out to roughly $0.16.

About the page fee. Google Search is heavily bot-protected and every page fetch consumes a paid proxy request whether or not it returns results. The $0.004 page fee covers that; a page that comes back empty costs only the page fee, never the per-image fee.

Set a maximum cost per run and the Actor stops cleanly when it reaches it. Proxy and compute are included — you are never billed separately for usage.

Input

{
  "queries": ["eiffel tower at night", "brutalist architecture"],
  "maxImagesPerQuery": 100,
  "imageSize": "large",
  "imageType": "photo",
  "safeSearch": "moderate",
  "country": "us",
  "language": "en"
}

Field	Description
queries	Search terms, one per line. Batch them — one run with 20 queries beats 20 runs.
maxImagesPerQuery	1–500. Above 100 fetches extra pages, each billed as one page request.
safeSearch	off, moderate or strict
imageSize	any, large, medium or icon
imageType	any, photo, clipart, lineart, gif or face
licence	any, creative_commons or commercial
country / language	Google gl and hl codes
proxyConfiguration	Google SERP proxy is on by default. Leave it on.
requestDelaySeconds	Pause between requests (default 2)

Output

{
  "type": "image",
  "query": "eiffel tower at night",
  "page": 1,
  "title": "Eiffel Tower at night - Wikipedia",
  "image_url": "https://upload.wikimedia.org/wikipedia/commons/a/a8/Tour_Eiffel.jpg",
  "thumbnail_url": "https://encrypted-tbn0.gstatic.com/images?q=tbn:ABC",
  "source_url": "https://en.wikipedia.org/wiki/Eiffel_Tower",
  "source": "Google Images",
  "width": 3024,
  "height": 4032,
  "search_url": "https://www.google.com/search?q=eiffel+tower+at+night&udm=2&hl=en&gl=us"
}

Queries that return nothing are written as {"type": "empty", ...} and failures as {"type": "error", ...}, so one bad query never kills the run.

Common uses

• Build image datasets for training or evaluation
• Source reference imagery for design and moodboards
• Find where an image appears across the web via source_url
• Bulk-collect product or location photography
• Give an AI agent real image URLs instead of hallucinated ones (this Actor supports agentic payments via x402 and Skyfire)

Integrations

Run on a schedule, or send output to Google Sheets, Airtable, Slack, Make, Zapier or your own webhook. Everything is reachable through the Apify API and the JavaScript and Python clients.

Notes and limits

• This is an unofficial scraper. Google changes its embedded result structure regularly; the parser uses a structural strategy with a regex fallback, and when a page yields nothing the Actor logs it and moves on rather than failing the run.
• Keep the proxy on. Google Search blocks unproxied traffic within a handful of requests.
• title and source_url are populated for most but not all results — see On partial fields above. image_url and dimensions are always present.
• thumbnail_url is usually null: Google inlines thumbnails into the page rather than serving them as URLs. Use image_url and resize on your side if you need small versions.
• The licence filter reflects Google's own labelling, which is a hint rather than a guarantee. Always verify usage rights at the source before reusing an image.
• Image URLs point at third-party servers and can go dead or change without notice.
• Scrape only public data and use the output in line with Google's terms, applicable law, and the copyright of each image's owner.

Support

Found a bug or need a field that is not extracted yet? Open an issue on the Actor's Issues tab.