Create a new image from an existing Deep Learning VM instance
Stay organized with collections
Save and categorize content based on your preferences.
Installing NVIDIA drivers on a new VM instance can be
time consuming, especially if you are creating numerous images. One way to
prevent this is to create your own image that is based on one of the
Deep Learning VM images, but that already has the NVIDIA drivers
preinstalled.
This topic describes how to create a new image based on an existing
Deep Learning VM image.
Create a new instance
First, follow the instructions in one of the following topics to create a new
instance. Be sure to include at least one GPU in your new instance.
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Hard to understand","hardToUnderstand","thumb-down"],["Incorrect information or sample code","incorrectInformationOrSampleCode","thumb-down"],["Missing the information/samples I need","missingTheInformationSamplesINeed","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2025-09-04 UTC."],[[["\u003cp\u003eThis guide outlines the process of creating a custom Deep Learning VM image with pre-installed NVIDIA drivers to save time on future setups.\u003c/p\u003e\n"],["\u003cp\u003eYou'll start by creating a new Deep Learning VM instance that includes at least one GPU, following one of the provided guides.\u003c/p\u003e\n"],["\u003cp\u003eAfter ensuring the NVIDIA drivers are installed on the instance, you will then stop the instance to create a custom image from it.\u003c/p\u003e\n"],["\u003cp\u003eThe final step is running a command to generate the custom image, complete with pre-installed NVIDIA drivers, allowing for quicker deployment of new Deep Learning VM instances.\u003c/p\u003e\n"]]],[],null,["# Create a new image from an existing Deep Learning VM instance\n\nInstalling NVIDIA drivers on a new VM instance can be\ntime consuming, especially if you are creating numerous images. One way to\nprevent this is to create your own image that is based on one of the\nDeep Learning VM images, but that already has the NVIDIA drivers\npreinstalled.\n\nThis topic describes how to create a new image based on an existing\nDeep Learning VM image.\n\nCreate a new instance\n---------------------\n\nFirst, follow the instructions in one of the following topics to create a new\ninstance. Be sure to include at least one GPU in your new instance.\n\n- [Creating a TensorFlow Deep Learning VM\n Instance](/deep-learning-vm/docs/tensorflow_start_instance)\n- [Creating a PyTorch Deep Learning VM\n Instance](/deep-learning-vm/docs/pytorch_start_instance)\n- [Creating a Deep Learning VM Instance from the Command\n Line](/deep-learning-vm/docs/cli)\n- [Creating a Deep Learning VM Instance from the Cloud\n Marketplace](/deep-learning-vm/docs/cloud-marketplace)\n\nVerify NVIDIA driver install\n----------------------------\n\nOnce the instance has booted, verify that the NVIDIA driver has been\ninstalled:\n\n1. SSH to your image: \n\n ```\n gcloud compute ssh \"DEPLOYMENT_NAME\"\n ```\n2. Run the following command: \n\n ```\n nvidia-smi\n ```\n\nIf the drivers have been installed, you can continue.\n\nStop the instance\n-----------------\n\nStop the instance by running the following command: \n\n```\ngcloud compute instances stop \"DEPLOYMENT_NAME\"\n```\n\nCreate your own image\n---------------------\n\nNow you create your own image based on the stopped instance. Run the following\nat the command line, giving the new image a name and a family name: \n\n```\ngcloud compute images create \"NEW_IMAGE_NAME\" \\\n --source-disk DEPLOYMENT_NAME \\\n --source-disk-zone ZONE \\\n --family NEW_FAMILY_NAME\n```\n\nOnce the command is finished running, you have a new image with NVIDIA drivers\npreinstalled that you can use to create new Deep Learning VM instances."]]
