Social Video Downloader

Paste a public link from YouTube, Instagram, Facebook, or X. View the title, author, and thumbnail first, then download the video or the MP3 audio. No sign-up. No app install.

YouTube — Video, MP3, ShortsInstagram — Reels, Posts, PhotosFacebook — Watch, ReelsX — Video, GIF, Photos
How downloads work here: This page first tries your own server — a small Python + yt-dlp server you run yourself (see "Your own server" below). It returns the title, author, description, tags and thumbnail for YouTube, Instagram, Facebook and X, and makes the video (MP4, up to 1080p) and the MP3 song itself with ffmpeg. If your own server is not set up yet, the page falls back to a Cloudflare Worker proxy with public Cobalt engines. Every attempt is shown in the log below the form.

Paste a link

The platform is detected automatically.

Run the server on your phone in Termux (setup below), expose it with a Cloudflare Tunnel, and paste the tunnel URL here. It is saved on this device only. When set, every download goes through your own server — no other downloader service is used.

1. Checking the link…
2. Fetching title, author, and thumbnail…
3. Requesting a download link from your own server…
Technical details

—

Advanced — yt-dlp command (PC / Termux)

Test links

Public test links from the research. Tap one to fill the form. If a post was deleted or made private, use Share → Copy link on a fresh public post instead.

What you actually get, by platform

Based on research and live checks in November 2026. No false promises — this is what each platform allows this page to return.

YouTube

  • Details: Title, author, and thumbnail on this page. Description and tags also, via the Invidious public API when it is available.
  • Video: 144p to maximum quality, including Shorts and Music, on a working Cobalt instance.
  • MP3: Yes, audio mode, up to 320 kbps.
  • Limitation: YouTube is disabled on the official Cobalt instance. A self-hosted instance is the reliable route.

Instagram

  • Details: Your own server returns the title and author for Instagram. The fallback engines do not return Instagram metadata.
  • Video / Photo: Reels, posts, and carousels. Carousels open as a picker, one item at a time.
  • MP3: Yes, where the reel has audio.
  • Limitation: Public posts only. Private accounts will fail.

Facebook

  • Details: Your own server returns the title and author for Facebook. The fallback engines do not return Facebook metadata.
  • Video: Public Watch videos and Reels, HD / SD where available.
  • MP3: Yes, with your own server — it converts the audio to MP3 with ffmpeg.
  • Limitation: Facebook often shows a login wall. The link must open while logged out.

X (Twitter)

  • Details: Post text, author, and thumbnail via the FxTwitter API.
  • Video: Yes, via Cobalt and via direct MP4 variants from FxTwitter.
  • MP3 / GIF: Yes, audio mode; GIF conversion is available in Cobalt.
  • Limitation: Cobalt notes X support may not be 100% reliable. Multi-media posts open as a picker.

How it works

Three steps, no storage. Nothing is saved on this site.

1. Details first

YouTube details come from oEmbed via Noembed, plus description and tags from Invidious. X details come from FxTwitter. Instagram and Facebook do not expose metadata to this page, so full details there need yt-dlp on a server or in Termux.

2. Cobalt creates the file link

The page sends a POST request to a Cobalt API instance. A tunnel or redirect response is the actual file. MP3 uses downloadMode “audio”. Carousels return a picker.

3. Nothing is stored

Cobalt streams the file to your device and does not keep it. This page keeps no history on a server.

Reliable setup — Cloudflare + your own Cobalt instance

Public engines can stop working at any time. The recommended setup is your own server: a small Python program (server.py) that uses yt-dlp directly — no other downloader service, no copying another site. It runs on your phone in Termux and is exposed with a Cloudflare Tunnel. Cloudflare Pages and Workers cannot run yt-dlp or ffmpeg, so the server has to run on a device you control.

Your own server — Termux setup (recommended)

1. Copy server.py and setup_termux.sh to your phone (Termux home folder, in a folder named own-downloader).

2. In Termux run: bash setup_termux.sh — it installs Python, ffmpeg and cloudflared, installs yt-dlp, starts the server on port 8080, and starts a Cloudflare Tunnel.

3. Copy the tunnel URL it prints (https://….trycloudflare.com) and paste it into the "Your own server" box above, then tap Save.

After that, the site uses only your server: title, author, description, tags and thumbnail come from it, Download video gives a merged MP4 (up to 1080p), and MP3 Audio gives a 192 kbps MP3 made with ffmpeg on your phone. The server never stores files — each download is made fresh and deleted after sending.

1. Run a Cobalt instance (Docker)

Create a docker-compose.yml file. Only API_URL is required:

services:
  cobalt:
    image: ghcr.io/imputnet/cobalt:11
    environment:
      - API_URL=https://api.your-domain.com/
      - CORS_WILDCARD=1
    ports:
      - "9000:9000"
    restart: unless-stopped

Then run docker compose up -d. Protect a public instance with Turnstile or an API key. Source: github.com/imputnet/cobalt — docs/run-an-instance.md

2. Cloudflare Worker proxy

The browser cannot always call another site's API directly (CORS), and an API key must never be in page code. Use a Pages Function / Worker in between. The page then calls only its own /api/download endpoint.

// functions/api/download.js
export async function onRequestPost({ request, env }) {
  const { url, downloadMode, videoQuality, audioFormat, audioBitrate, filenameStyle } = await request.json();
  const r = await fetch(env.COBALT_URL + "/", {
    method: "POST",
    headers: { "Accept": "application/json", "Content-Type": "application/json",
      ...(env.COBALT_KEY ? { "Authorization": "Api-Key " + env.COBALT_KEY } : {}) },
    body: JSON.stringify({ url, downloadMode, videoQuality, audioFormat, audioBitrate, filenameStyle })
  });
  return new Response(await r.text(), { headers: { "Content-Type": "application/json" } });
}

Store COBALT_URL and the key in Cloudflare Dashboard → Settings → Variables / Secrets.

3. Deploy on Cloudflare Pages

Dashboard → Workers & Pages → Create → Pages → Connect to Git, or use Direct Upload: npx wrangler pages deploy ./dist --project-name=your-site. Free tier: static requests and bandwidth are unlimited, 500 builds per month.

4. Full-power method — yt-dlp (Termux / PC)

yt-dlp cannot run in a browser or on Cloudflare (it needs Python and ffmpeg), but in Termux or on a PC it returns title, description, tags, duration, and views.

# Details only (JSON):
yt-dlp -J "LINK"
# Video, up to 1080p:
yt-dlp -f "bestvideo[height<=1080]+bestaudio/best" "LINK"
# MP3 only, best quality:
yt-dlp -x --audio-format mp3 --audio-quality 0 "LINK"

Version used in research: yt-dlp stable 2026.10.26, Python 3.11+. Run yt-dlp -U to update it.