cobalt

What Is cobalt?

cobalt is a video downloader that supports multiple video sites.

WHOIS records didn't reveal a domain registration date, but URLert and the Internet Archive both point to a September 2023 launch — making it relatively new in the video downloader space.

cobalt Is Open Source

cobalt's defining feature is that it's open-source software (OSS). The full codebase is publicly available here:

https://github.com/imputnet/cobalt

This means you can download cobalt and run it locally on your own device. The advantages of running it locally include not sending any data to cobalt's servers (more secure) and potentially faster downloads.

You can also deploy it to any server of your choice to avoid the overhead of running it locally. And since the code is fully accessible, you can use AI to modify it — for example, to add support for video sites cobalt doesn't currently handle.

You Can Send Questions and Feature Requests to cobalt

Through GitHub (the platform where cobalt's code is hosted), anyone can submit questions, feature requests, and bug reports via the Issues and Discussions pages. Engineers engage there frequently.

GitHub is primarily an engineering platform, but cobalt appears to welcome input from non-technical users as well. Creating a GitHub account is free, so joining the cobalt community is straightforward.

How to Use cobalt

There's a URL input field on the homepage. Paste the URL of the video page you want to download and submit — if cobalt supports that site, the download starts automatically.

If you submit a URL from an unsupported site, cobalt displays this message:

couldn't find any media that i could download for you. are you sure you pasted the right link?

How cobalt Downloads Videos

The approach varies by site, but cobalt generally works by retrieving direct video links and downloading from them.

The per-site processing logic is split into separate files and can be reviewed here:

https://github.com/imputnet/cobalt/tree/main/api/src/processing/services

The general flow appears to be: parse the submitted video page URL, extract the page content, and retrieve any authentication credentials needed to access the direct link — since most direct video links can't simply be fetched with a plain GET request.

When video and audio are delivered as separate streams, cobalt fetches each independently and merges them using ffmpeg.

Things to Watch Out For

According to cobalt's homepage, the following sites are supported:

bilibili, bluesky, dailymotion, facebook, instagram, loom, ok, pinterest, newgrounds, reddit, rutube, snapchat, soundcloud, streamable, tiktok, tumblr, twitch clips, twitter, vimeo, vk

That said, some of these sites didn't work during testing.

Because cobalt uses site-specific logic rather than a generic approach, any change a video site makes to its implementation can break downloads for that site. Compared to other downloaders, cobalt is more likely to silently stop working for a given site without warning.