How to use Docker Model Runner with Promptrak

Published: 2026-02-02 · #Promptrak

This guide explains how to use the Docker Model Runner with Promptrak so you can run models in containers and drive them from Promptrak no matter where you are.

Prerequisites

Enable and Configure Docker Model Runner

  1. With Docker Desktop open, click the Settings button. Doing so will open the Settings view.

Docker Desktop Default View

  1. With the settings view open, click the AI option in the left navigation. Doing so will show the Docker Model Runner settings. Configure its settings to match the screenshot below.

    • Enable Docker Model Runner by clicking the checkbox.
    • Enable Host-side TCP support. We set our port to 12434.
    • Set CORS to Custom and add the following origin https://promptrak.com.
    • Click the APPLY and CLOSE buttons.

Docker Desktop AI Settings for Docker Model Runner

  1. With the settings view closed, click the Models option in the left navigation. Doing so will open the Models view.

    • Click the BROWSE MODELS button. Doing so will show the Models search view.

Docker Desktop Models View

  1. With the Models Search view open, search for a model and pull its container. We’re going to use Gemma3 because of its smaller size.

    • Search for Gemma3 by typing it into the Search field. Doing so will filter the list of models.
    • Click the PULL text located within the Gemma3 model card to begin downloading the models container. Doing so will update the models view with a download progress bar.

Docker Desktop Models Search

  1. Wait for Docker Desktop to download the Gemm3 model.

Docker Dekstop Models Download

  1. Once downloaded, click the Run button. Doing so will show the Model Chat view.

Docker Desktop Running Model

  1. Enter a query and hit the Submit button.

    • Enter What AI Model are you? into the Ask a question… input.

Docker Desktop Querying a Model

  1. You should see a response similar to the one we received in the screenshot below.

Docker Desktop Viewing Query Result

Getting this far means you’ve successfully configured Docker Model Runner and have queried your AI docker model through chat. Next we’ll configure Promptrak to use this model!

Create and Configure Promptrak

  1. With Promptrak opened, open the Account Menu and click Settings. Doing so will open the Settings view.

Promptrak Landing Promptrak Account Menu 2. Under Local models click Add Runner button. Doing so will open the Add Runner dialog.

  1. Configure your Local Model Runner

    • Click the ADD button. Doing so will save the Local Model Runner configuration and add it to the Local Models list.

  2. Run your model in Docker (e.g. start the container that serves the model API). Note the host and port you expose (e.g. localhost:8080).

  3. In Promptrak, add or select a model runner of type “Docker Model Runner” (or equivalent name in your Promptrak version).

  4. Configure the runner with the Docker service URL (e.g. http://localhost:8080/v1 or whatever path your image uses) and any auth or model-name settings.

  5. Select the model in Promptrak so it matches the model served by the container.

  6. Run prompts from Promptrak; they are sent to the Dockerized service and responses appear in Promptrak.

Exact setup (image name, env vars, port) depends on the Docker image you use. See Promptrak’s docs for the Docker runner’s required fields and options. If you get connection errors, verify the container is up, the port is correct, and the API path matches what Promptrak is calling.