Endpoints for inference in production

APPLIES TO: Azure CLI ml extension v2 (current) Python SDK azure-ai-ml v2 (current)

After you train machine learning models or pipelines, you need to deploy them to production so that others can use them for inference. Inference is the process of applying new input data to the machine learning model or pipeline to generate outputs. While these outputs are typically referred to as "predictions," inferencing can be used to generate outputs for other machine learning tasks, such as classification and clustering. In Azure Machine Learning, you perform inferencing by using endpoints and deployments. Endpoints and deployments allow you to decouple the interface of your production workload from the implementation that serves it.


Suppose you're working on an application that predicts the type and color of a car, given its photo. For this application, a user with certain credentials makes an HTTP request to a URL and provides a picture of a car as part of the request. In return, the user gets a response that includes the type and color of the car as string values. In this scenario, the URL serves as an endpoint.

A diagram showing the concept of an endpoint.

Furthermore, say that a data scientist, Alice, is working on implementing the application. Alice knows a lot about TensorFlow and decides to implement the model using a Keras sequential classifier with a RestNet architecture from the TensorFlow Hub. After testing the model, Alice is happy with its results and decides to use the model to solve the car prediction problem. The model is large in size and requires 8 GB of memory with 4 cores to run. In this scenario, Alice's model and the resources, such as the code and the compute, that are required to run the model make up a deployment under the endpoint.

A diagram showing the concept of a deployment.

Finally, let's imagine that after a couple of months, the organization discovers that the application performs poorly on images with less than ideal illumination conditions. Bob, another data scientist, knows a lot about data augmentation techniques that help a model build robustness on that factor. However, Bob feels more comfortable using Torch to implement the model and trains a new model with Torch. Bob wants to try this model in production gradually until the organization is ready to retire the old model. The new model also shows better performance when deployed to GPU, so the deployment needs to include a GPU. In this scenario, Bob's model and the resources, such as the code and the compute, that are required to run the model make up another deployment under the same endpoint.

A diagram showing the concept of an endpoint with multiple deployments.

Endpoints and deployments

An endpoint is a stable and durable URL that can be used to request or invoke a model. You provide the required inputs to the endpoint and get the outputs back. An endpoint provides:

  • a stable and durable URL (like endpoint-name.region.inference.ml.azure.com),
  • an authentication mechanism, and
  • an authorization mechanism.

A deployment is a set of resources and computes required for hosting the model or component that does the actual inferencing. A single endpoint can contain multiple deployments. These deployments can host independent assets and consume different resources based on the needs of the assets. Endpoints have a routing mechanism that can direct requests to specific deployments in the endpoint.

To function properly, each endpoint must have at least one deployment. Endpoints and deployments are independent Azure Resource Manager resources that appear in the Azure portal.

Online and batch endpoints

Azure Machine Learning allows you to implement online endpoints and batch endpoints. Online endpoints are designed for real-time inference—when you invoke the endpoint, the results are returned in the endpoint's response. Batch endpoints, on the other hand, are designed for long-running batch inference. Each time you invoke a batch endpoint you generate a batch job that performs the actual work.

When to use online vs batch endpoint for your use-case

Use online endpoints to operationalize models for real-time inference in synchronous low-latency requests. We recommend using them when:

  • You have low-latency requirements.
  • Your model can answer the request in a relatively short amount of time.
  • Your model's inputs fit on the HTTP payload of the request.
  • You need to scale up in terms of number of requests.

Use batch endpoints to operationalize models or pipelines for long-running asynchronous inference. We recommend using them when:

  • You have expensive models or pipelines that require a longer time to run.
  • You want to operationalize machine learning pipelines and reuse components.
  • You need to perform inference over large amounts of data that are distributed in multiple files.
  • You don't have low latency requirements.
  • Your model's inputs are stored in a storage account or in an Azure Machine Learning data asset.
  • You can take advantage of parallelization.

Comparison of online and batch endpoints

Both online and batch endpoints are based on the idea of endpoints and deployments, which help you transition easily from one to the other. However, when moving from one to another, there are some differences that are important to take into account. Some of these differences are due to the nature of the work:


The following table shows a summary of the different features available to online and batch endpoints.

Feature Online Endpoints Batch endpoints
Stable invocation URL Yes Yes
Support for multiple deployments Yes Yes
Deployment's routing Traffic split Switch to default
Mirror traffic for safe rollout Yes No
Swagger support Yes No
Authentication Key and token Microsoft Entra ID
Private network support Yes Yes
Managed network isolation Yes Yes (see required additional configuration)
Customer-managed keys Yes Yes
Cost basis None None


The following table shows a summary of the different features available to online and batch endpoints at the deployment level. These concepts apply to each deployment under the endpoint.

Feature Online Endpoints Batch endpoints
Deployment types Models Models and Pipeline components
MLflow model deployment Yes Yes
Custom model deployment Yes, with scoring script Yes, with scoring script
Model package deployment 1 Yes (preview) No
Inference server 2 - Azure Machine Learning Inferencing Server
- Triton
- Custom (using BYOC)
Batch Inference
Compute resource consumed Instances or granular resources Cluster instances
Compute type Managed compute and Kubernetes Managed compute and Kubernetes
Low-priority compute No Yes
Scaling compute to zero No Yes
Autoscaling compute3 Yes, based on resources' load Yes, based on job count
Overcapacity management Throttling Queuing
Cost basis4 Per deployment: compute instances running Per job: compute instanced consumed in the job (capped to the maximum number of instances of the cluster).
Local testing of deployments Yes No

1 Deploying MLflow models to endpoints without outbound internet connectivity or private networks requires packaging the model first.

2 Inference server refers to the serving technology that takes requests, processes them, and creates responses. The inference server also dictates the format of the input and the expected outputs.

3 Autoscaling is the ability to dynamically scale up or scale down the deployment's allocated resources based on its load. Online and batch deployments use different strategies for autoscaling. While online deployments scale up and down based on the resource utilization (like CPU, memory, requests, etc.), batch endpoints scale up or down based on the number of jobs created.

4 Both online and batch deployments charge by the resources consumed. In online deployments, resources are provisioned at deployment time. However, in batch deployment, no resources are consumed at deployment time but when the job runs. Hence, there is no cost associated with the deployment itself. Notice that queued jobs do not consume resources either.

Developer interfaces

Endpoints are designed to help organizations operationalize production-level workloads in Azure Machine Learning. Endpoints are robust and scalable resources and they provide the best of the capabilities to implement MLOps workflows.

You can create and manage batch and online endpoints with multiple developer tools:

  • The Azure CLI and the Python SDK
  • Azure Resource Manager/REST API
  • Azure Machine Learning studio web portal
  • Azure portal (IT/Admin)
  • Support for CI/CD MLOps pipelines using the Azure CLI interface & REST/ARM interfaces

Next steps