Deploy Stable Diffusion on Koyeb
Learn more about Koyeb
·
Explore the documentation
·
Discover our tutorials
Koyeb is a developer-friendly serverless platform to deploy apps globally. No-ops, servers, or infrastructure management.
This repository is designed to show how to deploy Stable Diffusion to Koyeb. The Dockerfile
allows for configuration through environment variables to make deployment and configuration more straightforward.
Follow the steps below to deploy Stable Diffusion to your Koyeb account.
To use this repository, you need:
- A Koyeb account to build the
Dockerfile
and deploy it to the platform. If you don't already have an account, you can sign-up for free. - Access to GPU Instances on Koyeb.
The fastest way to deploy Stable Diffusion is to click the Deploy to Koyeb button below.
Clicking on this button brings you to the Koyeb App creation page with most of the settings pre-configured to launch this application.
Additionally, open the Health checks section and set the Grace period to 300 seconds to allow time for vLLM to fetch the model.
To modify this application example, you will need to fork this repository. Checkout the fork and deploy instructions.
If you want to customize and enhance this application, you need to fork this repository.
If you used the Deploy to Koyeb button, you can simply link your service to your forked repository to be able to push changes. Alternatively, you can manually create the application as described below.
On the Koyeb Control Panel, on the Overview tab, click the Create Web Service button to begin.
- Select GitHub as the deployment method.
- Choose the repository containing your application code.
- In the Instance section, select the GPU category and choose RTX-4000-SFF-ADA.
- In the Health checks section, set the Grace period to 300 seconds. This will provide time for Stable Diffusion to download the appropriate model from Hugging Face and initialize the server.
- Click Deploy.
The repository will be pulled, built, and deployed on Koyeb. Once the deployment is complete, it will be accessible using the Koyeb subdomain for your service.
If you have any questions, ideas or suggestions regarding this application sample, feel free to open an issue or fork this repository and open a pull request.