OpenCut is an MIT-licensed, self-hostable video editor branding itself as the open source CapCut alternative, but the trending GitHub repo isn't the version you should install today.
Every marketing team cutting short-form video for social eventually hits the same wall. Someone needs a CapCut Pro seat to strip the watermark, unlock the better export presets, or use the AI cleanup tools that make an edit look less like a first pass. That's a subscription that scales with headcount: one editor, one login, multiplied across however many people on the team touch video. OpenCut wants to delete that line item. It's an MIT-licensed video editor sitting at 84,700 stars and 8,300 forks on GitHub, and its own tagline calls it, flatly, "the open source CapCut alternative."
That's not a small claim, and it's not vaporware either. The repo has 96 contributors, a tagged release history, and enough momentum today to land on GitHub's trending list. The pitch is straightforward: a free, self-hostable video editor for web, desktop, and mobile that you run on your own infrastructure instead of renting a seat on someone else's. Self-hosted software can't raise your price mid-contract, and it can't get acquired and reprice the way commercial editing tools have a habit of doing.
Here's where it gets more interesting, and more honest, than the star count suggests. OpenCut is being rewritten from the ground up right now. The roadmap is genuinely ambitious: an editor API, a plugin-first architecture for third-party extensions, one Rust core shared across desktop, mobile, and browser, an MCP server so AI agents can drive the editor, and a headless mode for batch rendering. That's a real product vision. It is also, by the maintainers' own admission in the README, not what you should install if you need a working editor this week. They say directly that the previous build, kept at opencut-app/opencut-classic, is "the one to reach for today," and that opencut.app itself still runs that classic version while the rewrite lives at a separate subdomain until it's ready.
There's a smaller tell in the same direction. OpenCut's own landing page still advertises "70k+ stars," a number the GitHub repo passed a while ago on its way to 84,700. Even the project's marketing copy hasn't caught up to its own GitHub momentum, which is a minor thing but a useful gut check before taking any of the positioning at face value.
For a marketing ops lead or a content-heavy SaaS company, here's the actual translation. MIT licensing means you can self-host the editor, modify it, and never renegotiate a per-seat contract again. A team with even modest engineering support, someone who can stand up a small service and keep it patched, could run the classic build behind the company's own login instead of paying per editor per month for a commercial tool. That's real leverage against a subscription that grows every time you hire another person who needs to trim a reel. It's the same logic that made self-hosted alternatives attractive in scheduling, support, and analytics: the software is free, the leverage comes from not being billed per seat forever.
The honest caveat matters here more than usual. We tried to verify CapCut's current Pro pricing directly to put a number next to the savings, and the page didn't load, so no specific dollar figure appears in this article. Treat any number you see elsewhere with the same skepticism. The bigger caveat is what "self-hostable" is actually asking of you today: the classic build needs a real install, with a Rust toolchain, the proto and moon tooling, and a team member who can babysit updates, and the actively developed repo says plainly it isn't taking outside contributions "while the architecture is being designed." Self-hosting isn't free. It costs the hours of whoever sets it up and keeps it running, and right now those hours go toward a project mid-transition rather than a finished product.
None of that means the story is hype. It means the story is a bet, and 84,700 people have made it, on a team that's rewriting its entire architecture in public and telling you exactly which version not to trust yet. If you're evaluating this for your own team, the move isn't to install the trending repo. It's to bookmark opencut-classic, watch how the rewrite lands, and remember that the README that tells you not to use it yet is more trustworthy than the one that would have said nothing at all.