YouTube Video Downloader — MP4, MP3 & Subtitles [Pay Per Download]

Download YouTube videos and Shorts as MP4 (360p up to 4K) or extract MP3 audio — in bulk, by URL or video ID. Every download comes with full metadata, a shareable file link, and optional VTT subtitles. You are only charged for successful downloads — failures cost nothing.

Built for content teams, archivists, AI/ML pipelines, and anyone who needs YouTube media as actual files instead of links that expire.

Why Use This Downloader?

• Residential proxy included, not billed to you — YouTube blocks datacenter IPs, so every download needs residential. Ours runs on our own pool, so your bill carries no proxy line at all. Measured platform usage is about $0.0035 per video, whatever its length
• Works out of the box — that proxy is on by default, so your first run doesn't die on YouTube's "sign in to confirm you're not a bot" wall
• Pay per download, never for failures — a video that can't be downloaded is a free error row, not a charge
• Any link format — watch URLs, Shorts, embeds, live-VOD links, youtu.be short links, YouTube Music URLs, or bare 11-character video IDs
• Quality control — pick 360p, 480p, 720p, 1080p, 1440p, 2160p 4K, or best available
• MP3 audio mode — extract audio only; smaller, faster, ideal for transcription pipelines
• Subtitles included — save captions or auto-generated subtitles as clean VTT files with their own links
• Metadata-only mode — pre-check a list of videos (title, duration, views) for a fraction of a cent each before committing to downloads
• Bulk-ready — hundreds of links per run, duplicates removed automatically, 1–5 parallel downloads
• Age-restricted support — supply your own YouTube cookies to unlock age-restricted or members-only videos
• Smart fallback chain — residential proxy → alternate player clients → mobile endpoint, then a direct connection as a last resort, automatically per video
• Never hangs, never a surprise bill — every download has a hard timeout, so one stuck video can't freeze the run or quietly run up cost; it's killed, billed as a free error, and the run moves on. You pay a flat fee per successful video, so the most a run can cost is capped before you start it
• Deliver to your own bucket — optionally upload every file straight to your own S3, DigitalOcean Spaces, Cloudflare R2, Wasabi, Backblaze B2, or MinIO bucket; the returned link points at your storage, not Apify's

Coming from another YouTube downloader?

The complaints users file against YouTube download actors, and how this one is built to avoid them:

The complaint elsewhere	What this Actor does instead
"Sign in to confirm you're not a bot"	Residential proxy is on by default — the bot wall only blocks datacenter IPs, so a stock run gets through. For age-restricted or members-only videos there's a youtubeCookies field on top.
"Most videos not downloading / just won't download"	A per-video fallback chain (residential proxy → alternate player client → mobile endpoint, then direct as a last resort) retries each video several different ways before giving up, on a different exit IP each time. And when a video genuinely can't be fetched, it costs $0 — failures are never charged.
"The run gets stuck at 100% / runs forever / racked up a huge surprise bill"	Every download attempt has a hard timeout (perVideoTimeoutSecs, default 300s). A stuck video is killed, marked as a free error, and the run moves on — so one bad video can't freeze the run for hours. Cost is a flat fee per successful video, never per megabyte and never per retry, so a run's maximum charge is bounded before you press Start.
"Not getting an MP4 / not the format I asked for"	Output is explicit: MP4 at your chosen resolution cap (video+audio merged with ffmpeg into one playable file) or MP3 audio. What you select is what lands in the row, named in quality/format fields.
"The run 'succeeded' but I got nothing"	A run that downloads zero videos is reported as a failure, loudly — never a green run wrapping an empty dataset.
"Surprise per-MB fees on saved files"	Event pricing is a flat fee per successful video ($0.02–0.05 by quality, $0.02 MP3). The residential proxy is included — it runs on our own pool, not billed to you, so there is no per-GB proxy line on your bill. Measured platform usage is about $0.0035 per video.
"Direct download links stopped working"	Files are stored in your run's key-value store with stable HTTPS links that survive as long as your storage retention — not IP-locked CDN URLs that expire in hours.

Overview

This actor takes a list of YouTube video URLs or IDs and produces one dataset row per video containing the video's metadata (title, channel, duration, views, upload date), the download details (quality, format, file size), and stable links to the downloaded files stored in the run's key-value store.

It is a downloader, not a search tool — you supply the video links. To discover videos first (by search query, channel, or Shorts feed), pair it with the scrapers listed in Explore More Scrapers; their output plugs directly into this actor's input.

Supported Inputs

Input type	Example
Watch URL	https://www.youtube.com/watch?v=jNQXAC9IVRw
Shorts URL	https://www.youtube.com/shorts/IbkzTfJCFjk
Short link	https://youtu.be/z9rdrNrkvDY
Embed / live URL	https://www.youtube.com/embed/jNQXAC9IVRw
YouTube Music URL	https://music.youtube.com/watch?v=jNQXAC9IVRw
Bare video ID	jNQXAC9IVRw

Extra URL parameters (?si=, &t=10s, ?feature=share) are handled automatically. Duplicate videos in the list are downloaded only once.

Not supported: playlist URLs, channel URLs, and search-result URLs (feed those to the companion scrapers first and pass the resulting video URLs here). Live streams that are currently airing cannot be downloaded until they finish.

Use Cases

Who	What for
Content & social media teams	Archive campaign videos, repurpose content across platforms
AI / ML engineers	Build audio/video datasets; MP3 mode feeds Whisper-style transcription directly
Journalists & researchers	Preserve evidence and sources before videos disappear
Marketing agencies	Competitor creative analysis, ad-library building
Educators	Offline copies of reference material
Media monitoring	Automated capture of brand mentions in video form

How It Works

1. Paste video links — any mix of watch/Shorts/youtu.be URLs or bare IDs, one per line.
2. Pick quality and format — MP4 at your chosen resolution cap, or MP3 audio-only.
3. The engine downloads each video through a battle-tested fallback chain (residential proxy → alternate player clients → mobile endpoint, with a direct connection as the last resort), merging streams with ffmpeg.
4. Files are stored in the run's key-value store with stable, shareable HTTPS links.
5. One row per video lands in the dataset — metadata, file size, download link, subtitles link — exportable as JSON, CSV, or Excel.

Input Configuration

Field	Type	Default	Description
videoUrls	array	— (required)	YouTube video URLs (incl. YouTube Music) or bare 11-char video IDs. Free Apify plans process the first 3 per run
quality	string	720p	Max resolution: 360p, 480p, 720p, 1080p, 1440p, 2160p, best. Ignored for MP3
format	string	mp4	mp4 (video) or mp3 (audio-only extraction)
downloadSubtitles	boolean	false	Also save subtitles/auto-captions as a VTT file
subtitleLanguages	array	["en"]	Subtitle language codes in priority order — the first one the video actually has is the one saved (en matches en-US), and subtitlesLanguage reports which
transcriptionAndSubtitle	boolean	false	Subtitles plus a plain-text transcript built from the caption file (transcriptUrl + inline transcriptText). Reads YouTube's existing captions — a video with captions disabled returns no transcript, there is no speech recognition here. Billed as one subtitles download
metadataOnly	boolean	false	Skip the media download, return only metadata rows. Caption files are the exception: with a subtitle toggle on they are still fetched, stored and charged
youtubeCookies	string	—	cookies.txt content (Netscape format) for age-restricted videos
proxy	object	residential	On by default — it's what gets past YouTube's datacenter-IP block. Set {"useApifyProxy": false} only if you bring your own proxies or cookies
maxConcurrency	integer	2	Parallel downloads (1–5)
perVideoTimeoutSecs	integer	300	Max seconds for a single download attempt before it's killed and the next strategy is tried. Stops a stuck video from hanging the whole run; a timed-out video is a free error row. Also bounds the upload of the finished file, with a 15-minute floor — raise it for multi-GB 4K transfers
filenameTemplateParts	array	[]	Compose the stored filename from title, videoId, channel, date, quality. Parts are joined in the order the array holds them — the Console records the order you tick them; rearrange on the JSON tab. Empty keeps the current video-<videoId>-<timestamp>.mp4 naming
storageType	string	apify	Where files go: apify (this run's key-value store), s3, azure, or gcs — fill in only that backend's fields below
storeInKVStore	boolean	false	When files go to your own bucket, also keep a copy in the key-value store (kvStoreUrl, kvStoreSubtitlesUrl, kvStoreTranscriptUrl). A failed mirror never affects the copy in your own bucket
s3Bucket	string	—	Target bucket. Required when storageType is s3
s3Region	string	auto	Bucket region, e.g. eu-central-1. AWS rejects a bucket addressed with the wrong region, so left empty the Actor asks S3 where the bucket is and uses the answer. Ignored when s3Endpoint is set
s3Endpoint	string	—	Leave empty for AWS S3. For S3-compatible providers set the endpoint, e.g. https://nyc3.digitaloceanspaces.com (Spaces) or https://<account>.r2.cloudflarestorage.com (R2)
s3AccessKeyId	string (secret)	—	Access key for the bucket
s3SecretAccessKey	string (secret)	—	Secret key for the bucket
s3Path	string	—	Optional key prefix, e.g. youtube/exports. Used by all three cloud backends, not just S3
s3Acl	string	—	Optional canned ACL, e.g. public-read to make objects publicly downloadable. S3 only — Azure and GCS ignore it and return 7-day signed links instead
azureConnectionString	string (secret)	—	Azure storage-account connection string. Required when storageType is azure. Links come back signed with a 7-day read-only SAS when the string carries an account key
azureContainerName	string	—	Target container (3–63 chars, lowercase, single hyphens). Required when storageType is azure
googleCloudBucketName	string	—	Target bucket. Required when storageType is gcs. Links come back as 7-day V4 signed read URLs — no ACL is set
googleCloudServiceKey	string (secret)	—	Service-account key JSON (raw or base64). Required when storageType is gcs; needs Storage Object Creator on the bucket

Typical download run (proxy defaults to residential — nothing to configure):

{
    "videoUrls": [
        "https://www.youtube.com/watch?v=jNQXAC9IVRw",
        "https://youtu.be/z9rdrNrkvDY"
    ],
    "quality": "720p",
    "format": "mp4",
    "downloadSubtitles": true
}

Cheap pre-check before a big batch:

{
    "videoUrls": ["jNQXAC9IVRw", "z9rdrNrkvDY", "IbkzTfJCFjk"],
    "metadataOnly": true
}

Audio for a transcription pipeline:

{
    "videoUrls": ["https://www.youtube.com/watch?v=jNQXAC9IVRw"],
    "format": "mp3"
}

Deliver files straight to your own S3-compatible bucket:

{
    "videoUrls": ["https://www.youtube.com/watch?v=jNQXAC9IVRw"],
    "quality": "1080p",
    "storageType": "s3",
    "s3Bucket": "my-media-bucket",
    "s3Region": "us-east-1",
    "s3AccessKeyId": "AKIA...",
    "s3SecretAccessKey": "...",
    "s3Path": "youtube",
    "s3Acl": "public-read"
}

Pricing — What a Download Really Costs

This actor uses pay-per-event + platform usage pricing. Two components appear on your bill, and we'd rather you understand both up front:

1. Event fees (paid to the developer) — flat per successful download: $0.02–0.05 per video depending on quality, $0.02 per MP3, $0.005 per subtitle file, $0.002 per metadata row. Failed downloads are never charged.

2. Platform usage (paid to Apify at your plan's rates) — compute and data transfer only. Residential proxy is NOT on this bill: YouTube blocks datacenter IPs, so every download goes through a residential pool, but it is our own pool and its bandwidth is not passed on to you. There is no Apify RESIDENTIAL line on a run of this actor.

   Measured on a real run (4 videos, 61 MB, build 0.0.21): $0.0141 total — $0.0106 compute, $0.0030 internal data transfer, $0.0002 external. That is about $0.0035 of platform usage per video.

So a typical 720p download costs about $0.03 in event fees + ~$0.0035 in platform usage — the event fee is essentially the whole bill, and it is fixed per video regardless of length. A 1-hour podcast costs the same $0.03 as a 3-minute clip. If you have unused monthly platform credits, the usage side consumes those first. You can also cap any run with Apify's maximum-charge limit.

Output Overview

Each video produces exactly one dataset row: identity (videoId, url), metadata (title, channel, channelUrl, durationSec, uploadDate, viewCount), download details (quality, format, fileSizeBytes), file links (downloadUrl, subtitlesUrl, subtitlesLanguage), and storageTarget, which names where the files actually went. The transcript and key-value-store-mirror columns appear only when you switch those features on, so a default run's rows carry no permanently empty columns. Failed videos produce the same row shape with an error field and null file links — and no charge.

Files live in the run's key-value store by default; the downloadUrl is a direct HTTPS link you can open, share, or feed to another system. Point storageType at your own S3, Azure or Google Cloud bucket and downloadUrl becomes a link to that bucket instead. A download that succeeds but cannot be delivered at all is reported with storageError and the Storage upload failed: prefix on error, so a misconfigured bucket never looks like a missing video — and it is not charged. When only part of the delivery fails — a mirror copy, or a caption file after the video itself landed — storageError says so on its own and error stays unset, because the file is there and the row must not read as empty.

Output Samples

Successful 720p download with subtitles (from a real run):

{
    "videoId": "jNQXAC9IVRw",
    "url": "https://www.youtube.com/watch?v=jNQXAC9IVRw",
    "title": "Me at the zoo",
    "channel": "jawed",
    "channelUrl": "https://www.youtube.com/channel/UC4QobU6STFB0P71PMvOGN5A",
    "durationSec": 19,
    "uploadDate": "2005-04-24",
    "viewCount": 399449655,
    "quality": "720p",
    "format": "mp4",
    "fileSizeBytes": 533916,
    "downloadUrl": "https://api.apify.com/v2/key-value-stores/…/records/video-jNQXAC9IVRw-1784027309504.mp4",
    "subtitlesUrl": "https://api.apify.com/v2/key-value-stores/…/records/subtitles-jNQXAC9IVRw-1784027309504.vtt",
    "subtitlesLanguage": "en"
}

Metadata-only row (metadataOnly: true — no file, $0.002):

{
    "videoId": "jNQXAC9IVRw",
    "title": "Me at the zoo",
    "channel": "jawed",
    "durationSec": 19,
    "viewCount": 399449655,
    "uploadDate": "2005-04-24",
    "downloadUrl": null
}

Key Output Fields

Group	Fields
Identity	videoId, url
Video metadata	title, channel, channelUrl, durationSec, uploadDate, viewCount
Download details	quality, format, fileSizeBytes
File links	downloadUrl, subtitlesUrl, subtitlesLanguage
Transcript	transcriptUrl, transcriptSource (captions, auto-captions, or null when no metadata said which), transcriptText (inline, truncated at 20k chars) — present only when transcriptionAndSubtitle is on
Delivery	storageTarget, plus kvStoreUrl / kvStoreSubtitlesUrl / kvStoreTranscriptUrl when storeInKVStore is on
Diagnostics	error (nothing was delivered — every file link is null and the row is not charged), storageError (delivery trouble: beside error when nothing landed, on its own when the main file landed and a mirror or caption file did not)

FAQ

Which URLs work? Individual video links only — watch, Shorts, embed, youtu.be, or bare IDs. Playlists and channels are not expanded; use the companion channel/search scrapers to get video URLs first.

Why is residential proxy on by default? YouTube blocks requests from datacenter IP ranges with a "sign in to confirm you're not a bot" wall, so every proxied attempt runs first and a direct connection is kept only as a last resort, for the case where the proxy itself is unavailable. Each attempt is pinned to a single exit IP, because YouTube ties a media URL to the address that requested it and a rotating proxy makes the file download 403 after the metadata already succeeded. The proxy runs on our own residential pool and is not billed to you (see Pricing above); to run without it entirely, set "proxy": {"useApifyProxy": false} and ideally supply youtubeCookies.

How long do the file links live? Files sit in the run's key-value store and follow your account's storage retention (default 7+ days depending on plan). Download or transfer files you want to keep long-term.

Can it download age-restricted or members-only videos? Supply youtubeCookies (exported from a logged-in browser session in Netscape format) and the actor passes them to the download engine. Members-only content requires a membership on the supplying account.

What about currently-live streams? Not supported while live. Once the stream ends and becomes a regular video, it downloads normally.

Does it charge for failed videos? No. Failures produce an error row and zero event charges, and the proxy bandwidth a failed attempt burned is on us, not on you.

Is 4K supported? Yes — set quality to 2160p (or 1440p for QHD, or best for the highest stream the video offers). Note that 4K files are large (often 500 MB+), which matters for proxy usage costs.

Support

• For issues or feature requests, please use the Issues section of this actor.
• If you need customization or have questions, feel free to contact the author:
  • Author's website: https://muhamed-didovic.github.io/
  • Email: muhamed.didovic@gmail.com

Additional Services

Need something tailored? Custom output formats, delivery to your S3/GCS bucket, scheduled bulk archiving, integration with your transcription pipeline, or a private variant of this actor — all possible. Get in touch via the contacts above.

Explore More Scrapers

Pair this downloader with the rest of the YouTube toolkit — their output feeds straight into videoUrls:

• YouTube Search & Video Metrics Scraper — find videos by search query, get analytics-ready rows
• YouTube Shorts Scraper — pull Shorts feeds by channel
• YouTube Comments Scraper — comments with replies for any video
• YouTube Channel Videos Details Scraper — full channel video inventories
• Video Transcripts MCP Server — plain-text transcripts for AI agents
• YouTube Transcript Scraper with Whisper AI Fallback — SRT/VTT/JSON transcripts with translation

🤖 For AI Agents & LLM Apps

Compact reference for AI agents calling this actor via the Apify MCP server or the Apify API (actor: memo23/youtube-video-downloader).

Purpose: Download individual YouTube videos/Shorts as MP4 (360p–4K) or MP3 audio by URL or ID, with metadata, VTT subtitles, and shareable file links; only successful downloads are charged.

Minimal input (residential proxy is applied by default — no proxy field needed):

{
  "videoUrls": ["https://www.youtube.com/watch?v=jNQXAC9IVRw"]
}

Cheap pre-check variant: { "videoUrls": ["jNQXAC9IVRw"], "metadataOnly": true } returns metadata rows only, no download.

Output: one row per video — videoId, url, title, channel, channelUrl, durationSec, uploadDate, viewCount, quality, format, fileSizeBytes, downloadUrl, subtitlesUrl, subtitlesLanguage, storageTarget, plus transcriptUrl/transcriptSource/transcriptText when the transcript toggle is on; failed videos carry error and null file links.

Behaviors an agent should know:

• There is no maxItems field — the length of videoUrls is the effective cap; duplicates are deduped. Keep lists bounded since you pay per successful download.
• Individual videos only — playlist, channel, and search URLs are not expanded, and currently-live streams can't be downloaded until they finish.
• Residential proxy is on by default (YouTube blocks datacenter IPs); pass "proxy": {"useApifyProxy": false} only when supplying own proxies or cookies. Quality accepts 360p–2160p or best; YouTube Music and live-VOD URLs are valid inputs.
• metadataOnly: true is a cheap pre-check ($0.002/row); format: "mp3" is ~10× cheaper than video; youtubeCookies (Netscape format) unlocks age-restricted/members-only videos.
• Billing: per-event to the developer — $0.02–0.05/video, $0.02/MP3, $0.005/subtitle, $0.002/metadata row; failures never charged. Apify platform usage is compute and data transfer only, measured at ~$0.0035 per video; residential proxy is included and never billed to the caller.

---

⚠️ Disclaimer

This actor is an independent tool and is not affiliated with, endorsed by, or sponsored by YouTube or Google LLC. "YouTube" and related trademarks are the property of Google LLC and are used here solely to describe the tool's function.

The actor processes only publicly accessible content. You are responsible for how you use the downloaded material: respect YouTube's Terms of Service, copyright law, and the content creators' rights in your jurisdiction. Downloading copyrighted content for redistribution without permission is illegal in most countries. Use of this tool for personal archiving, research, and other fair-use purposes is the intended scope. Comply with applicable data-protection regulations (GDPR, CCPA) where relevant.

SEO Keywords

youtube video downloader, download youtube videos, youtube mp4 downloader, youtube to mp3, youtube shorts downloader, bulk youtube downloader, youtube downloader api, download youtube subtitles, youtube video to mp4, youtube audio extractor, youtube downloader for developers, youtube archive tool, download youtube videos in bulk, youtube video backup, yt video download api, youtube content archiving, youtube dataset builder, video download automation, youtube media pipeline, download youtube shorts mp4, youtube music downloader, 4k youtube downloader, youtube downloader no sign in bot check