For an art installation, this is the main repo.
It's deployed with Runpod and was impressively easy in contrast to your typical big clouds provider. I straight used Runpods image as base in my Dockerfile, knowing it works.
docker buildx build --platform linux/amd64 -t yourusername/sdxlturbo-api:latest .
docker login
docker push yourusername/sdxlturbo-api:latestor just the one I pushed:
causeri3/sdxlturbo-api:latest
- choose the default RTX3090 Pod
- in the settings:
- put in the Docker image
- expose port 8000
- sweet: once its running, you can click on connect - it straight gives you a public url
- optional: in your profile you can upload your public ssh key and it will automatically assign it to your pod
In case it helps anyone to make a decision, for comparison, the current setup:
| Seconds | GPU | RAM | GPU computing frameworks |
|---|---|---|---|
| 24 | RTX3090 | 125 GB | cuda |
| 210 | M3 Pro | 36 GB | mps |
Powered by Stability AI
This API uses the Stability AI model SDXL Turbo under the Stability AI Community License.