Introduction to Docker Offload

A step-by-step guide for getting started with Docker Offload

Docker announced Docker Offload last week, a new managed service that seamlessly integrates the local containers development workflow with a cloud-powered environment. This feature is designed to support AI/ML developers who require access to GPUs for builds and compute-intensive workloads like running LLMs, machine learning pipelines, and GPU-accelerated applications.

Note that this service comes with 300 free GPU minutes. The usage price after credits expire is currently $0.015 per GPU minute subject to change after Beta. Please check Docker website for the up-to-date pricing.

What Docker Offload Is

In a nutshell, Docker Offload enables you to seamlessly execute your builds and run containers in a fully managed cloud service from your local machine. Or in other words, it can extend your local development workflow to a scalable cloud environment when your local resources are not sufficient.

Key features:

• Easy to use (!!!)

• Provides access to NVIDIA L4 GPUs

• Execute docker build commands on powerful cloud infrastructure

• Run containers with cloud compute and GPU acceleration

• Same Docker commands, cloud execution

Getting Started with Docker Offload

To get started with Docker Offload, you will need:

• Docker account

• Docker Desktop v4.43.0 and above

There are two methods to connect to the cloud with Docker Offload:

• Docker Dashboard

• CLI

Connecting via Docker Dashboard

Docker Offload is currently under beta (July 2025). Therefore, you will have to enable it first from the settings dashboard by selecting the Beta features menu (marked in yellow), and check the Enable Docker Offload option (marked in green):

Likewise, in the case you need GPU support, you will have to enable it (marked in light blue).

Once you have enabled the feature, you should see on the Docker dashboard a toggle button to start a session:

After clicking the button, the icon changes to a cloud, and the dashboard top bar color changes from blue to deep purple:

Connecting via the CLI

Likewise, you can start a session using the CLI. To check the Docker Offload arguments via the CLI use the docker offload command:

docker offload

This should return the following:

Usage:  docker offload COMMAND

Docker Offload

Commands:
  accounts       Prints available accounts
  diagnose       Print diagnostic information
  start          Start a Docker Offload session
  status         Show the status of the Docker Offload connection
  stop           Stop a Docker Offload session
  version        Prints the version

Run 'docker offload COMMAND --help' for more information on a command.

The start command enables SSH to the cloud and starts a session:

docker offload start

This will prompt the following screen, which enables you to select the account you want to use for the session:

Next, it will prompt you to select whether you want GPU support or not:

Once you connect to the cloud, you should receive a connection confirmation:

What is amazing about this feature is that once you are connected, you don’t need to change anything else in your workflow. Any Docker command that you run from your local terminal will get executed on the cloud.

Using the context command, you can see that the Docker endpoint is now set to the cloud account:

docker context ls

If you are using multiple accounts, you can use the accounts command to get a list of accounts and their status:

docker offload accounts                                                                                                 

Since I have a single account, this will return the following JSON:

{
  "user": {
    "id": "xxxxxxx-xxxx-xxxxx",
    "fullName": "",
    "gravatarUrl": "",
    "username": "rkrispin",
    "state": "READY"
  },
  "orgs": []
}

Once you start a cloud session, the local images are not available and you either can pull them from your registery (e.g., Docker Hub) or rebuild them.

To end a session, use the docker offload stop command:

docker offload stop

Resources

• Docker Offload sign-up for beta page: https://www.docker.com/products/docker-offload/

• Docker Offload documentation: https://docs.docker.com/offload/

---