Why two calls instead of one?

Because a YouTube rendition is work, not a link that is already lying there. info costs one call and tells you the exact size of every quality; create builds the one you picked. Returning eight links from the first call would mean either building eight files or handing you links that break.

Is there a limit_mb parameter?

No, and there will not be. limit_mb exists for Telegram bots that must stay under an upload ceiling. Here you get file_size in bytes for every rendition before you commit, so the choice is yours.

How long does the download link live?

It carries a signed expiry (expires, a Unix timestamp) and answers 410 after it. Call create again for a fresh one. The link streams the file rather than serving a stored copy, so there is no Range support and no Content-Length — the size you plan against is file_size, which info reported before you chose.

Yes for every rendition except MP3, which we transcode and therefore estimate; the exact flag marks the difference. Video and M4A sizes come straight from the source.

YouTube Downloader API

This is the only two-step platform, and the reason is honest: on every other source the renditions already exist, so handing you all their links costs us nothing. A YouTube rendition does not exist until it is built — video and audio are separate streams that have to be muxed — so listing eight ready links would be a promise we could not keep. Instead info tells you exactly what each rendition weighs, and create builds the one you choose and returns a signed link with an expiry. The link is served from our side because YouTube signs its own media URLs to the IP that requested them; passed to your server they answer 403.

videoaudioGET /v1/youtube/info2 credits

Links we accept

Any public post URL on these hosts. Short links and country mirrors are resolved for you.

https://www.youtube.com/watch?v=dQw4w9WgXcQ
https://youtu.be/dQw4w9WgXcQ
https://www.youtube.com/shorts/abc123
https://music.youtube.com/watch?v=abc

Hosts

youtube.com · youtu.be · youtube-nocookie.com

Request

Step 1 — list the renditions (2 credits)

curl -s "https://api.saveapi.org/v1/youtube/info" \
  -H "Authorization: Bearer sk_live_xxxx" \
  -G --data-urlencode "url=https://www.youtube.com/watch?v=dQw4w9WgXcQ"

Step 2 — build the one you want (10 credits)

curl -s "https://api.saveapi.org/v1/youtube/create" \
  -H "Authorization: Bearer sk_live_xxxx" \
  -G --data-urlencode "url=https://www.youtube.com/watch?v=dQw4w9WgXcQ" \
  --data-urlencode "quality=720p"

Response

{
  "success": true,
  "platform": "youtube",
  "source_url": "https://youtu.be/dQw4w9WgXcQ",
  "video_id": "dQw4w9WgXcQ",
  "title": "Rick Astley - Never Gonna Give You Up (Official Video)",
  "author": "Rick Astley",
  "duration_seconds": 213,
  "duration_str": "3:33",
  "thumbnail": "https://i.ytimg.com/vi/dQw4w9WgXcQ/hqdefault.jpg",
  "is_short": false,
  "formats": [
    { "quality": "360p",  "file_size": 11330415, "file_size_str": "10.8 MB", "exact": true },
    { "quality": "720p",  "file_size": 29884416, "file_size_str": "28.5 MB", "exact": true },
    { "quality": "1080p", "file_size": 80546201, "file_size_str": "76.8 MB", "exact": true }
  ],
  "audio_formats": [
    { "quality": "m4a", "file_size": 3460300, "file_size_str": "3.3 MB", "exact": true },
    { "quality": "mp3", "file_size": 8520000, "file_size_str": "8.1 MB", "exact": false }
  ],
  "credits": { "spent": 2, "remaining": 99998 }
}

Questions before you build

Why two calls instead of one?: Because a YouTube rendition is work, not a link that is already lying there. info costs one call and tells you the exact size of every quality; create builds the one you picked. Returning eight links from the first call would mean either building eight files or handing you links that break.
Is there a limit_mb parameter?: No, and there will not be. limit_mb exists for Telegram bots that must stay under an upload ceiling. Here you get file_size in bytes for every rendition before you commit, so the choice is yours.
How long does the download link live?: It carries a signed expiry (expires, a Unix timestamp) and answers 410 after it. Call create again for a fresh one. The link streams the file rather than serving a stored copy, so there is no Range support and no Content-Length — the size you plan against is file_size, which info reported before you chose.
Are the sizes exact?: Yes for every rendition except MP3, which we transcode and therefore estimate; the exact flag marks the difference. Video and M4A sizes come straight from the source.

YouTube to MP3 API — audio link from any video or Short YouTube Shorts download API — MP4 from a Shorts link