Website Screenshot & PDF API: Full Page, Mobile, Element, Thumbnail

Give it a list of public URLs and get back a screenshot or PDF of each page, saved in the run's storage, plus one clean dataset row per URL with the file link, final URL, HTTP status, image size and timings. Cookie and consent banners are hidden without clicking anything, so captures show the page, not the pop-up. Failed, blocked and timed-out URLs are never charged.

Built for:

• AI agents and LLM pipelines that need to "see" a page. Turn on the free thumbnail for a small JPEG sized for vision models instead of downloading a large PNG.
• SEO and marketing agencies making audit decks: desktop and mobile homepage captures for a whole client list in one run.
• QA and release teams archiving how pages looked, before and after a deploy or migration.
• Monitoring and brand-safety tools that keep a visual record of landing pages and link targets.
• No-code users on Make, Zapier or n8n who want an image link for every URL in a sheet.

What it does

• Four capture types: viewport (what a visitor sees first), full page (the whole scrolling page, with lazy images loaded), element (one CSS selector, e.g. a pricing table) and PDF (A4, Letter and more, screen or print styling).
• Devices: desktop 1280×800, desktop HD 1920×1080, tablet 768×1024 at 2x, mobile 390×844 at 3x with a real mobile layout, or any custom size.
• Formats: PNG, JPEG or WebP with adjustable quality.
• Cookie banners hidden, never clicked: about 30 common consent managers plus a heuristic for home-grown banners. Nothing is accepted or rejected on anyone's behalf. Each row says whether a banner was hidden and flags full-screen consent walls it left in place.
• No idling: "network idle" waits are capped at 8 seconds and every URL has a hard time budget, so one chatty page can't stall or inflate a run.
• Honest rows: every URL gets a row. Failures carry a clear error code and are free.
• Run summary: a SUMMARY record with counts, an error histogram, average timings and total bytes, so automations can branch on one record.

Quick start

1. Paste URLs into URLs to capture, one per line.
2. Pick a Capture type and Device (defaults: viewport, desktop).
3. Click Start. Open the Gallery view of the results to see every capture; the files are in the run's key-value store.

Sample output

One dataset row for https://example.com with thumbnail on:

{
  "index": 0,
  "url": "https://example.com/",
  "finalUrl": "https://example.com/",
  "redirected": false,
  "ok": true,
  "statusCode": 200,
  "error": null,
  "errorDetail": null,
  "titlePresent": true,
  "captureType": "viewport",
  "device": "desktop",
  "viewport": { "width": 1280, "height": 800, "deviceScaleFactor": 1, "isMobile": false },
  "format": "png",
  "screenshotUrl": "https://api.apify.com/v2/key-value-stores/<storeId>/records/shot-0000-example-com-viewport.png",
  "imageUrl": "https://api.apify.com/v2/key-value-stores/<storeId>/records/shot-0000-example-com-viewport.png",
  "thumbnailUrl": "https://api.apify.com/v2/key-value-stores/<storeId>/records/thumb-0000-example-com-viewport.jpg",
  "image": { "key": "shot-0000-example-com-viewport.png", "url": "https://api.apify.com/v2/key-value-stores/<storeId>/records/shot-0000-example-com-viewport.png", "format": "png", "contentType": "image/png", "bytes": 21544, "width": 1280, "height": 800 },
  "thumbnail": { "key": "thumb-0000-example-com-viewport.jpg", "url": "https://api.apify.com/v2/key-value-stores/<storeId>/records/thumb-0000-example-com-viewport.jpg", "bytes": 9120, "width": 512, "height": 320 },
  "pdf": null,
  "pageHeight": 800,
  "truncated": false,
  "loadComplete": true,
  "consentHidden": false,
  "consentWall": false,
  "blockedPrivateRequests": 0,
  "timings": { "navigationMs": 640, "settleMs": 530, "captureMs": 180, "totalMs": 1420 },
  "chargedEvent": "screenshot",
  "capturedAt": "2026-10-24T12:00:00+00:00"
}

A failed URL (free; abbreviated):

{ "url": "https://blocked.example/", "ok": false, "error": "http_4xx_blocked", "errorDetail": "HTTP 403: the site refused the request; not retried.", "chargedEvent": null }

Error codes: invalid_url, blocked_host, private_network, robots_disallowed, http_4xx_blocked (401/403/429), challenge_page, timeout, navigation_failed, selector_not_found, browser_crash, capture_failed.

Inputs

Field	Default	What it does
URLs to capture	(required)	Public http(s) URLs, up to 500 per run.
Capture type	viewport	viewport, fullPage, element or pdf. One type per run.
Device	desktop	desktop, desktopHd, tablet, mobile or custom (width, height and pixel ratio fields).
Image format / quality	png / 80	png, jpeg or webp; quality applies to JPEG and WebP.
Element selector	—	CSS selector for the element type, e.g. #pricing.
Also make a thumbnail	off	Free JPEG preview, longest edge 128–1024 px (default 512).
Hide cookie banners	on	Hides consent pop-ups with CSS.
Also hide these elements	—	Your own selectors, e.g. chat widgets or sticky bars.
Dark mode	off	Requests the page's dark color scheme.
Wait until	load	load, domcontentloaded or networkidle (max 8 s extra).
Extra delay	500 ms	Settle time after load.
Wait for element	—	Waits for a selector; captures anyway if it never appears (loadComplete: false).
Scroll to load lazy images	on	Full page and PDF only.
Max height for full page	16,384 px	In CSS px, up to 30,000. Taller pages are cut and flagged truncated: true; the image limits under Limits can cut sooner.
PDF options	A4, portrait, backgrounds, screen	Paper size, landscape, backgrounds, screen or print styling.
Timeout per URL	30 s	Budget for loading, waiting and capturing one URL.
Pages in parallel	auto	One page per GB of memory (2 at 2 GB), up to 8.
Skip video and audio	on	Faster captures; images, fonts and CSS always load.

Pricing

Pay per event: you pay for captures that were saved, plus a small start fee per run (below).

Event	Free plan	Bronze	Silver	Gold
Screenshot (viewport, element or mobile; any format)	$0.003	$0.0025	$0.00225	$0.002
Full-page screenshot	$0.0045	$0.004	$0.0036	$0.0034
PDF	$0.0055	$0.005	$0.0045	$0.004

That's $2.00–$3.00 per 1,000 screenshots.

Start fee: each run also pays a start fee of $0.001 per GB of run memory, on every plan, because starting a browser costs more than one screenshot. At the default memory that is $0.001 for a run of 1–2 URLs (1 GB) and $0.002 for 3 or more URLs (2 GB); if you raise the memory yourself, it is $0.001 per GB (for example $0.004 at 4 GB). Batch URLs into one run to pay it once.

Thumbnails are free. Not charged: invalid or refused URLs, robots.txt blocks, 401/403/429 responses, bot-protection challenges, timeouts, missing selectors, browser crashes and anything that couldn't be saved. A rendered 404 or 500 page is captured and charged, because you get the image. Set Max cost per run to cap a run: it stops cleanly at the limit and the SUMMARY record lists the URLs it skipped.

Integrations and API

Call it from the Apify API or a client library with a list of URLs and read the dataset: each row's screenshotUrl is a direct link to the file (imageUrl is the same link, but empty for PDFs). Schedule it as a saved task for weekly site galleries, or connect it to Make, Zapier, n8n, Google Sheets or Slack through Apify's integrations. AI agents can call it through Apify's MCP server and use thumbnailUrl to look at pages cheaply.

Limits

• Public pages only. No logins, cookies, headers or session tokens are accepted, so pages behind a sign-in show whatever a signed-out visitor sees.
• Refused hosts: facebook.com, fb.com, instagram.com, linkedin.com, x.com, twitter.com, tiktok.com, accounts.google.com, login.microsoftonline.com, login.live.com and console.apify.com (login-walled or their terms forbid automated capture).
• Private and internal addresses are refused, including after redirects and DNS resolution: localhost, private networks, link-local and cloud metadata addresses. Requests a page makes to such addresses are blocked and counted in blockedPrivateRequests.
• robots.txt is always respected for the URL and for wherever it redirects (same host or another). A site that blocks bots in robots.txt gives a free robots_disallowed row.
• Bot protection is not bypassed. 401, 403 and 429 responses and challenge pages (for example "Just a moment...") are reported and not retried, and the run's remaining URLs on that host get free rows without being loaded. No proxies are used.
• Cookie banners are hidden, not dismissed. A full-screen consent wall is left in place and flagged consentWall: true (still charged, since the capture is delivered).
• Up to 500 URLs per run and 60 seconds per URL at most.
• Image size limits (full page and element captures are cut and flagged truncated: true at whichever comes first):
  • Max height, up to 30,000 CSS px.
  • 32,000 device pixels tall (WebP: 16,383). In CSS px that is 32,000 at 1x, 16,000 at 2x and about 10,666 at 3x (mobile).
  • 40 million device pixels in all, to keep memory bounded. For example about 31,250 CSS px tall at 1280 wide and 1x, 20,833 at 1920 and 1x, and 7,812 at 1280 and 2x.
  • Custom viewports wider than 1920 px use a pixel ratio of at most 2.
• Files live as long as your Apify storage retention for the run. Download or copy anything you need to keep.
• Video and audio don't play in captures; WebGL-heavy pages may render differently than in a desktop browser.

FAQ

Why is my screenshot missing the cookie banner? That's the point: it is hidden with CSS so you see the page. Turn off Hide cookie banners to capture it.

Why does a row say consentWall: true? The page covers most of the screen with a consent or paywall overlay. Hiding it could hide the page itself, so it is left in place and flagged.

The thumbnail of my full-page capture only shows the top of the page. Very tall pages are cropped to the top (up to twice as tall as wide) before shrinking, so the thumbnail stays readable. The full image is untouched.

My own site shows challenge_page or http_4xx_blocked. Allowlist the SaiberOpsBot User-Agent token in your firewall or CDN and run it again.

Can I get both a screenshot and a PDF? Run it twice with different capture types, or save two tasks.

Why is loadComplete false? The page's load event or your Wait for element didn't happen within the timeout. The page was captured anyway.

Responsible use and data sources

The only data sources are the URLs you submit and their robots.txt files. The browser identifies itself with Chrome's own User-Agent plus the token SaiberOpsBot, respects robots.txt (RFC 9309), never logs in, never solves CAPTCHAs and does not rotate IP addresses. Cookie banners are hidden on screen only: no consent is given, recorded or stored. Screenshots show whatever the page shows, which can include personal data on some pages; you are responsible for the URLs you submit and for how you use the captures. The consent selector list is our own.

Support

Found a bug or need a feature? Open an issue on this Actor's Issues tab and we'll respond there.

Maintained by Saiber Operations LLC.