OpenCreator: An Open AI Studio for Video Translation, Editing, and Generation

One subscription for video translation, a second for voiceover, a third for generating cover images. While an expert puts a clip together from three services, a competitor has already published theirs. There is an open project that gathers part of these tasks into one desktop app.

It's called OpenCreator, formerly known as KrillinAI. The repository is open, github.com/krillinai/OpenCreator, Apache 2.0 license, more than 12,400 stars on GitHub as of September 29, 2026, with updates almost every day. Next is a look at what is inside, based on the project documentation, without promises that aren't there.

What kind of tool it is

OpenCreator is built on top of Codex CLI, OpenAI's agent environment: instead of its own engine for the agent, the project adds a desktop app, templates, and visual tools for content creators around it. It runs as a desktop app for macOS and Windows, or as a web version that a developer launches from source.

Inside there are two linked modes: a visual desktop with ready content-creation tools, and a dialogue with the agent, where you describe the task in plain text and the agent picks the right tool itself and takes it to a result.

What is actually inside

According to the documentation as of September 29, 2026, the project has ten ready creator tools and two more in development:

• - video translation into 101 languages with bilingual subtitles or voiceover;
• - a video downloader from YouTube, TikTok, Instagram, Pinterest, and several other platforms;
• - a cover generator from a topic, a video link, and a reference;
• - image generation through GPT Image;
• - video generation through Seedance;
• - article writing from a topic, links, or documents;
• - posts for Little Red Book from a topic or source;
• - a short video script, segmented and ready for shooting;
• - stick-figure-style animation from text or from a YouTube video;
• - script voiceover with a choice of voice, pace, and intonation.

Two tools, editing short clips from a long video and a digital avatar for a talking head, are marked "in development" directly in the documentation and are not yet available. It's an honest detail that is easy to miss in a short retelling on social media.

What is really free and what is not

The code and the app themselves are distributed under the Apache 2.0 license, which is free and open in the literal sense. But the actual work with text, images, video, and voice goes through models: ChatGPT, DeepSeek, Qwen, GPT Image, Seedance, and others, depending on what is connected in the settings. On first launch the app checks whether you have a working ChatGPT login or an API key, and offers to use it.

In other words, the shell and the desktop around the agent are free. The cost of the generations themselves depends on which model provider you connect and on what terms you have access to it.

What you need to run it

For the desktop app on macOS or Windows you don't need to install Node.js or pnpm separately, since Codex CLI is already built into the installer. For the web version from source you need Node.js version 22 or newer, pnpm 9.15.0, a working Codex CLI, and a login to it. In both cases, real generation needs access to models, your own or through an already configured Codex.

Who it suits and who it doesn't

It suits people who already work with several AI services for content separately and want to bring some of the tasks, translation, subtitles, covers, script, into one desktop app instead of switching between tabs. The video translation tool for 101 languages is especially useful if part of the audience watches content not in Russian.

It doesn't suit those looking for a finished clip with one button, with no model setup and no digging in the agent environment. There is no digital avatar for a talking head here yet, and editing short clips from a long video is still in development.

I write about open content tools that I test for my own system in the channel «Растём с ИИ» ("Growing with AI"). A similar breakdown of a free tool for assembling video without a camera and editing, but with a different approach: MoneyPrinterTurbo, an honest test.

Bottom line

OpenCreator doesn't replace the whole bundle of services with one button, but it removes part of the switching between tabs: translation, subtitles, covers, and script are gathered in one place around an agent. The shell is free, and generations cost exactly as much as the connected model costs.

If you need not a bundle of separate tools but an assembled content system for your niche, you can start with a free consultation.