Download a video from Instagram or Reddit without visiting sketchy third-party downloader sites.
Extract audio from a SoundCloud track or video clip for offline listening.
Run your own private media downloader instance on your server without relying on hosted services.
Cobalt is a media downloader that lets you save videos, audio, and other content from social media platforms directly to your device. The problem it addresses is that most online media downloaders are cluttered with advertisements, trackers, pop-ups, paywalls, and dark patterns that make a simple download feel like an obstacle course. Cobalt aims to be the clean, no-nonsense alternative: paste a link, get the file, done. The way it works is by accepting a URL from a supported platform, processing the request through its API, and returning the media file directly to you. The project describes itself as working like a fancy proxy, it never caches any content, it just fetches and passes through what is publicly available. The platforms mentioned in the project topics include Instagram, Reddit, and SoundCloud, among others. There is no sign-in required, no account, and no tracking. The README notes that cobalt can only download free and publicly accessible content, and that the same content could be retrieved manually through a browser's developer tools. You would use cobalt if you want to save a video or audio clip from a social media platform without dealing with intrusive third-party downloader sites. You can use the hosted version at cobalt.tools, or you can run your own instance using the provided self-hosting documentation. The monorepo includes the API, the frontend web interface, and related packages. The tech stack is Svelte for the frontend web interface, with a JavaScript API backend, all licensed under AGPL-3.0.
Generated 2026-05-18 · Model: sonnet-4-6 · Verify against the repo before relying on details.