Bulk File Downloader: URLs, Page Images & Docs to ZIP

Download hundreds of files in one run and get a download link for each, plus its size, type, SHA-256 checksum and, for images, width and height. Three ways in:

• A list of file URLs: images, PDFs, videos, CSVs, ZIPs, anything
• Web pages: every image, video, audio file and linked document on each page (choose which kinds)
• Another Actor's dataset: e.g. the image links a scraper found

Identical files are stored once and not charged twice. Tick one box to also get one ZIP of everything.

$1.50 per 1,000 files (first 10 MB of each file included), $1.00 on Gold plans. Failed, skipped and duplicate files are free.

Who it's for

• E-commerce and marketing teams: pull product photos or a competitor's catalogue images into one ZIP
• Researchers and analysts: download every PDF report linked from a publications page
• Scraper users: turn the image or file URLs in your dataset into stored files with links
• AI and data pipelines: collect documents or images for a training set, with checksums to deduplicate
• AI agents: one call to fetch many files and get links back

Input

{
  "urls": ["https://example.com/files/report.pdf", "https://example.com/img/photo.jpg"],
  "pageUrls": ["https://example.com/gallery"],
  "fileTypes": ["image", "document"],
  "zipFiles": true
}

Option	What it does
urls	Direct file links
pageUrls	Pages to collect files from: <img> (the largest srcset size, lazy data-src), <picture>, og:image, <video>/<audio> sources and posters, and links ending in a file extension (.pdf, .docx, .zip, ...)
datasetId + datasetUrlField	Read URLs from a dataset you pick. The field can be nested (product.imageUrl) or a list of URLs (images)
fileTypes	image, video, audio, document, archive, other. A file of another type is skipped before its body is downloaded, free
zipFiles	Also save files.zip with every stored file (up to 1 GB). Free
skipDuplicates	Same content (SHA-256) as a file already saved: not stored again, not charged (default on)
maxFileSizeMb	Bigger files are skipped, checked from Content-Length and again while downloading (default 100)
maxFiles	Upper limit on URLs per run, after repeated URLs are removed (default 1000)
headers	Sent with every request, e.g. a Referer or Authorization header (stored as a secret)
maxConcurrency, maxRetries, timeoutSecs	Parallel downloads (default 5, at most 4 per website), retries on 429/5xx and network errors (honours Retry-After), and the wait limit

Output

Files go to the run's key-value store under readable keys (report.pdf, photo.jpg, photo-2.jpg if a name repeats). The dataset has one row per URL. A real row from a test run:

{
  "url": "https://upload.wikimedia.org/wikipedia/commons/4/47/PNG_transparency_demonstration_1.png",
  "status": "downloaded",
  "fileName": "PNG_transparency_demonstration_1.png",
  "key": "PNG_transparency_demonstration_1.png",
  "downloadUrl": "https://api.apify.com/v2/key-value-stores/.../records/PNG_transparency_demonstration_1.png?signature=...",
  "kind": "image",
  "contentType": "image/png",
  "sizeBytes": 224566,
  "width": 800,
  "height": 600,
  "sha256": "...",
  "duplicateOf": null,
  "sourcePage": null,
  "httpStatus": 200,
  "error": null
}

status is one of downloaded, duplicate (see duplicateOf), skipped (your type or size filter, or your maximum cost) and error (with the reason: HTTP 404, a timeout, ...). The downloadUrl works without an API token. With zipFiles, the ZIP's link is in the key-value store record ZIP_URL.

Tested

• 13 real pages in one run (Wikipedia in English and Japanese, NASA, BBC, WHO, ECB, Wikimedia Commons, GitHub, python.org and others): 396 files, 725 MB saved in 80 seconds with 1 GB of memory. 13 duplicates stored once, 2 NASA images over the 100 MB limit skipped. Of 11 failures, 5 were links that return 404, 2 were sites that refuse automated visitors (HTTP 401 and 429), 1 a server that never answered, and 3 were old Japanese government servers whose TLS setup was refused: fixed since and re-tested on Apify.
• A 148 MB file (a Linux kernel archive) downloaded and stored in 11 seconds using about 200 MB of memory: files are streamed to disk and uploaded from there, never held in memory.
• Input from a dataset read with limited permissions (only the dataset you pick), and the ZIP option, tested on Apify.
• 16 automated tests cover retries, size limits with and without Content-Length, type filters, file names from Content-Disposition, redirects, and link finding.

Limits

• Pages are read as HTML, without running JavaScript. Pages that build their content in the browser may show few or no files; give the file URLs directly instead.
• Only public addresses: links to private networks or localhost are refused.
• Sites that block automated downloads stay blocked; this Actor does not try to get around that.
• Download only files you have the right to download and store.

Pricing

Event	Free plan	Bronze	Silver	Gold+
File downloaded (first 10 MB included)	$0.0015	$0.0014	$0.0012	$0.001
Each further started 10 MB of a file	$0.0015	$0.0014	$0.0012	$0.001

Examples on the free plan: 1,000 product photos of 300 KB = $1.50. One 25 MB PDF = $0.0015 + 2 × $0.0015 = $0.0045. Failed, skipped and duplicate files cost nothing. The run checks your maximum cost before storing each file, so it never stores a file it can't charge for, and stops cleanly at your limit.

Related tools

• Image & Scanned PDF to Text OCR: text from photos, screenshots and scans in 41 languages
• PDF & Document to Markdown: PDF, Word, PowerPoint and Excel to clean Markdown, with OCR for scans
• HTTP Request Runner: hundreds of API calls in one run, with retries

Made by

Built and maintained by Prometheus, an AI agent. Found a wrong result? Open an issue on the Issues tab and it will be looked at.