Skip to content

Supported Images

On ScienceCloud, you don't have to install the operating system when you start a new virtual machine. Instead, we provide bootable disks ready to use, so that within one minute you can login to your new instance. These are called images, and are basically files containing bootable virtual disks. We provide some basic images, but you can also create your own and upload them to the cloud. One approach for building your own images is described in the Custom Images section.

Warning

Whenever you delete an instance, its root disk as well as ALL your customizations are removed. The base image, however, is never deleted.

Note that the image names below that being with three asterisks are the LATEST supported images. There may be other images available.

Table of supported images on the ScienceCloud

Image Name Default User Image ID
***Ubuntu 26.04 (2026-07-06) ubuntu 99c64133-8b5d-4f6a-854b-e869ee3ea11d
***Ubuntu 24.04 (2026-05-08) ubuntu a7b6cf57-335f-454a-be07-a5ae0b4d449a
***Debian 13.4 (2026-05-08) debian d359dee6-ab28-4fdf-ab12-c358fe9a8ae4
***Debian 12.13 (2026-05-08) debian 6c56c27f-88e7-4016-8f46-ba4938bc12ee
***Windows Server 2022 Standard (2025-05-15) de6c2733-8d31-4632-82a2-a0f59c88eff6

For further actions with images, check section Snapshots and Images.

Compared to the stock images, the following modifications have been made:

  • All Ubuntu images come with Apptainer pre-installed.
  • On Ubuntu 26, /tmp mounting as tmpfs is disabled because many scientific applications write to /tmp to reduce memory consumption. The behaviour is controlled via tmp.mount service. If you would prefer the default behaviour, i.e. mounting /tmp as tmpfs, you can unmask and re-enable the tmp.mount service.

GPU usage

Users of Ubuntu images should first install NVIDIA drivers, utilities like nvidia-smi, and cuda-toolkit with the following steps:

sudo apt update
sudo apt install ubuntu-drivers-common nvidia-cuda-toolkit -y
sudo ubuntu-drivers install
reboot

For more information about GPUs on ScienceCloud, see section GPU-enabled flavors.