Skip to content

Open-source AI video and image projects

People publish great AI tools for free, but most need a powerful computer and some setup. This page shows who made each one, and what you need to use it.

Runs here
Use it on this site with nothing to install. New accounts get free runs, then it costs credits.
Free download
We link to the author's repo. You install and run it yourself. We don't run it or charge for it.
Free for small teams
The same, but its licence is free only for individuals and small companies. Larger companies need a company licence.

Runs here

Only 2 of the projects we read could honestly run here. These use the author's prompt wording on the image model we already run, and make one image at a time. Your words and the result go through our safety checks.

Apps you can install

Each of these has its own setup, and you pay for any model or computer it uses.

Several were built for Comfy's Open Source Higgsfield challenge. Glidezo isn't connected to the challenge, to Comfy or to these authors.

Wan 2.2 and ComfyUI

Wan 2.2 makes video from text or an image on your own GPU, and ComfyUI is the usual way to run it.

Comfy publishes official Wan 2.2 workflows you can open in ComfyUI. Our Wan 2.2 page covers prices and other ways to run it.

  • Free download

    Wan 2.2

    Open video models. Comfy's guide says the series supports commercial use.

    You need: Your own GPU. Comfy's guide says the 5B version should fit on 8 GB with ComfyUI's offloading.

    By Wan-Video. Apache-2.0 licence.

  • Free download

    ComfyUI

    A visual node editor for image and video workflows. Many projects on this page run on it, and it opens the Wan workflows.

    You need: Your own computer with a GPU. Its README says it supports all operating systems and GPU types.

    By Comfy. GPL-3.0 licence.

Tools that make video from code

These need no AI model: you or your coding assistant write the video as code, and your computer renders it.

New to this? Start with our HyperFrames guide, which has free skills ready to use.

Questions

What is the difference between Runs here and Free download?
Runs here means you can use it on this site: type your idea, click, and get an image, with free runs for new accounts and credits after that. Free download means we only link to the author's repo. You install it yourself, and we don't run it, support it or charge for it.
Why do so few projects run here?
We don't rent GPUs. A project runs here only if what it does can honestly work on the image model we already use. Most of these projects need their own open models on a GPU, so a version here wouldn't be the same thing.
Is a run here the same as using the project itself?
No. A run uses the author's prompt wording on our image model, Nano Banana 2.1, and makes one image. Each skill page says what runs here and what only the free download does.
Can I use these commercially?
That's two separate questions. The code licence, shown on each project, covers the project's code. It says nothing about what a model makes, because each model has its own terms, so read those too. For Runs here, the model is Nano Banana 2.1 through a service called Sume. Sume's terms say it "does not claim ownership of generated outputs created for your account", and that "you are responsible for reviewing outputs before publication, distribution, or commercial use". Those terms are between Sume and Glidezo, and we can't promise that a result is cleared for what you want to do. For the free downloads, we have read Wan 2.2's terms through Comfy's guide, but not those of most other models.

Sources