Skip to main content

Deploy Your First Model

Make your first deployment quickly


Clarifai offers an intuitive interface and powerful APIs that streamline the process of deploying your machine learning models, allowing you to get up and running in no time.

With just a few simple steps or a minimal amount of code, you can take a trained model and set it up for inference.

Set up Your Account or Log in

Create a new Clarifai account or log into your existing one to start accessing the platform's powerful AI capabilities.

Step 1: Identify a Model

Log in to the Clarifai platform and identify the model you want to deploy on the homepage. Then, click the DEPLOY THE MODEL button in the bottom right corner of the model's information card.

Step 2: Review Your Compute Instances

After clicking the button, a small window will appear, displaying the pre-configured compute instances available for deployment. Review the options and choose the one that best fits your needs.

  • Basic Compute — Recommended for development and quick tests, offering reliable, low-cost performance.
  • Advanced Compute — Ideal for large-scale production inference or training of complex models.

tip

If you prefer more control and want to deploy the model using an existing cluster and nodepool, click the provided link in the pop-up window. This will allow you to customize the deployment based on your infrastructure.

Then, click the Deploy button.

A compute cluster and nodepool will be automatically created using the pre-configured settings. The model will be deployed within this infrastructure.

Step 3: Run Inferences

After clicking the button, you'll be redirected to the created nodepool page, where your compute settings and deployed model will be listed.

You can then run inferences by:

  • Navigating to the model’s individual page.
  • Selecting the deployment from the Deployment dropdown.
  • Going to the Overview pane and submitting your prediction requests.

Alternative Deployment Method

Alternatively, you can initiate your first deployment by navigating to the Set Up Compute section on the homepage. Then, choose either the Basic Compute or Advanced Compute option, as outlined earlier, to apply your pre-configured compute orchestration settings. If these predefined options don’t fully meet your requirements, you can select the Create your own option to customize the infrastructure according to your specific needs.

Once you’ve selected a pre-configured option, a confirmation window will appear, displaying your chosen compute settings. A Personal Access Token (PAT) will be pre-populated for your convenience. If needed, you can select a different token from the dropdown list or create a new one.

After confirming your settings, click the Create button. This will automatically generate a compute cluster and nodepool based on your pre-configured settings.

Next, choose the model you wish to deploy and click the Deploy Model button. You’ll then be redirected to the nodepool page, where your compute settings and deployed model will be displayed for reference, as outlined earlier.