Migrate your applications to Vertex AI

Vertex AI brings together AI Platform and legacy AutoML services under one unified UI and API to simplify the process of building, training, and deploying machine learning models. With Vertex AI, you can move from experimentation to production faster, efficiently discover patterns and anomalies, make better predictions and decisions, and stay agile in the face of changing priorities and market conditions. This page helps determine the changes you need to make when you migrate your applications from legacy AutoML or AI Platform to Vertex AI.

Vertex AI supports all features and models available in legacy AutoML and AI Platform. However, the client libraries do not support client integration backward compatibility. In other words, you must plan to migrate your resources to benefit from Vertex AI features.

This page compares the API methods used to complete common user journeys so that you can see how your project's applications could be updated to use the Vertex AI API.

Common user journeys

Select the tab for your product, and then click a user journey to see how the Vertex AI API methods compare with the API methods used by your existing applications.

Legacy AutoML Natural Language

Starting on September 15, 2024, you can only customize classification, entity extraction, and sentiment analysis objectives by moving to Vertex AI Gemini prompts and tuning. Training or updating models for Vertex AI AutoML for Text classification, entity extraction, and sentiment analysis objectives will no longer be available. You can continue using existing Vertex AI AutoML Text models until June 15, 2025. For a comparison of AutoML text and Gemini, see Gemini for AutoML text users. For more information about how Gemini offers enhanced user experience through improved prompting capabilities, see Introduction to tuning. To get started with tuning, see Model tuning for Gemini text models

Click one of the following user journeys:

• AutoML Natural Language: Train and deploy a text classification model

• AutoML Natural Language: Train and deploy a text entity extraction model

• AutoML Natural Language: Train and deploy a text sentiment model

Legacy AutoML Natural Language: Train and deploy a text classification model

To see an example of how to use Vertex AI to train and deploy a text classification model, run the "Vertex AI Migration: AutoML Text Classification" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the legacy AutoML API and the Vertex AI API, read about differences between the legacy AutoML Natural Language and Vertex AI products, and then use the following table to help migrate your API.

Step	Legacy AutoML Natural Language	Vertex AI
Create a dataset	projects.locations.datasets.create	projects.locations.datasets.create
	projects.locations.datasets.importData	projects.locations.datasets.import
Train a model	projects.locations.models.create	projects.locations.trainingPipelines.create
		projects.locations.trainingPipelines.get
Evaluate the model	projects.locations.models.modelEvaluations.list	projects.locations.models.evaluations.list
	projects.locations.models.modelEvaluations.get	projects.locations.models.evaluations.get
Make batch predictions	projects.locations.models.batchPredict	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get
Make online predictions	projects.locations.models.deploy	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
	projects.locations.models.predict	projects.locations.endpoints.predict

Back to top

Legacy AutoML Natural Language: Train and deploy a text entity extraction model

To see an example of how to use Vertex AI to train and deploy a text entity extraction model, run the "Vertex AI Migration: AutoML Text Entity Extraction" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the legacy AutoML API and the Vertex AI API, read about differences between the legacy AutoML Natural Language and Vertex AI products, and then use the following table to help migrate your API.

Step	Legacy AutoML Natural Language	Vertex AI
Create a dataset	projects.locations.datasets.create	projects.locations.datasets.create
	projects.locations.datasets.importData	projects.locations.datasets.import
Train a model	projects.locations.models.create	projects.locations.trainingPipelines.create
		projects.locations.trainingPipelines.get
Evaluate the model	projects.locations.models.modelEvaluations.list	projects.locations.models.evaluations.list
	projects.locations.models.modelEvaluations.get	projects.locations.models.evaluations.get
Make batch predictions	projects.locations.models.batchPredict	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get
Make online predictions	projects.locations.models.deploy	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
	projects.locations.models.predict	projects.locations.endpoints.predict

Back to top

Legacy AutoML Natural Language: Train and deploy a text sentiment model

To see an example of how to use Vertex AI to train and deploy a text sentiment model, run the "Vertex AI Migration: AutoML Text Sentiment Analysis" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the legacy AutoML API and the Vertex AI API, read about differences between the legacy AutoML Natural Language and Vertex AI products, and then use the following table to help migrate your API.

Step	Legacy AutoML Natural Language	Vertex AI
Create a dataset	projects.locations.datasets.create	projects.locations.datasets.create
	projects.locations.datasets.importData	projects.locations.datasets.import
Train a model	projects.locations.models.create	projects.locations.trainingPipelines.create
		projects.locations.trainingPipelines.get
Evaluate the model	projects.locations.models.modelEvaluations.list	projects.locations.models.evaluations.list
	projects.locations.models.modelEvaluations.get	projects.locations.models.evaluations.get
Make batch predictions	projects.locations.models.batchPredict	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get
Make online predictions	projects.locations.models.deploy	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
	projects.locations.models.predict	projects.locations.endpoints.predict

Back to top

Legacy AutoML Video Intelligence

Click one of the following user journeys:

• AutoML Video Intelligence: Train and deploy an object tracking model

• AutoML Video Intelligence: Train and deploy a video classification model

Legacy AutoML Video Intelligence: Train and deploy an object tracking model

To see an example of how to use Vertex AI to train and deploy an object tracking model, run the "Vertex AI Migration: AutoML Video Object Tracking" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the legacy AutoML API and the Vertex AI API, read about differences between the legacy AutoML Video and Vertex AI products, and then use the following table to help migrate your API.

Step	Legacy AutoML Video	Vertex AI
Create a dataset	projects.locations.datasets.create	projects.locations.datasets.create
	projects.locations.datasets.importData	projects.locations.datasets.import
Train a model	projects.locations.models.create	projects.locations.trainingPipelines.create
		projects.locations.trainingPipelines.get
Evaluate the model	projects.locations.models.modelEvaluations.list	projects.locations.models.evaluations.list
	projects.locations.models.modelEvaluations.get	projects.locations.models.evaluations.get
Make batch predictions	projects.locations.models.batchPredict	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get

Back to top

Legacy AutoML Video Intelligence: Train and deploy a video classification model

To see an example of how to use Vertex AI to train and deploy a video classification model, run the "Vertex AI Migration: AutoML Video Classification" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the legacy AutoML API and the Vertex AI API, read about differences between the legacy AutoML Video and Vertex AI products, and then use the following table to help migrate your API.

Step	Legacy AutoML Video	Vertex AI
Create a dataset	projects.locations.datasets.create	projects.locations.datasets.create
	projects.locations.datasets.importData	projects.locations.datasets.import
Train a model	projects.locations.models.create	projects.locations.trainingPipelines.create
		projects.locations.trainingPipelines.get
Evaluate the model	projects.locations.models.modelEvaluations.list	projects.locations.models.evaluations.list
	projects.locations.models.modelEvaluations.get	projects.locations.models.evaluations.get
Make batch predictions	projects.locations.models.batchPredict	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get

Back to top

Legacy AutoML Vision

Click one of the following user journeys:

• AutoML Vision: Train and deploy an image classification model

• AutoML Vision: Train and deploy an object detection model

Legacy AutoML Vision: Train and deploy an image classification model

To see an example of how to use Vertex AI to train and deploy an image classification model, run the "Vertex AI Migration: AutoML Image Classification" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the legacy AutoML API and the Vertex AI API, read about differences between the legacy AutoML Vision and Vertex AI products, and then use the following table to help migrate your API.

Step	Legacy AutoML Vision	Vertex AI
Create a dataset	projects.locations.datasets.create	projects.locations.datasets.create
	projects.locations.datasets.importData	projects.locations.datasets.import
Train a model	projects.locations.models.create	projects.locations.trainingPipelines.create
		projects.locations.trainingPipelines.get
Evaluate the model	projects.locations.models.modelEvaluations.list	projects.locations.models.evaluations.list
	projects.locations.models.modelEvaluations.get	projects.locations.models.evaluations.get
Make batch predictions	projects.locations.models.batchPredict	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get
Make online predictions	projects.locations.models.deploy	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
	projects.locations.models.predict	projects.locations.endpoints.predict
Train and export an Edge model	projects.locations.models.create	projects.locations.trainingPipelines.create
		projects.locations.trainingPipelines.get
	projects.locations.models.export	projects.locations.models.export

Back to top

Legacy AutoML Vision: Train and deploy an object detection model

To see an example of how to use Vertex AI to train and deploy an object detection model, run the "Vertex AI Migration: AutoML Image Object Detection" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the legacy AutoML API and the Vertex AI API, read about differences between the legacy AutoML Vision and Vertex AI products, and then use the following table to help migrate your API.

Step	Legacy AutoML Vision	Vertex AI
Create a dataset	projects.locations.datasets.create	projects.locations.datasets.create
	projects.locations.datasets.importData	projects.locations.datasets.import
Train a model	projects.locations.models.create	projects.locations.trainingPipelines.create
		projects.locations.trainingPipelines.get
Evaluate the model	projects.locations.models.modelEvaluations.list	projects.locations.models.evaluations.list
	projects.locations.models.modelEvaluations.get	projects.locations.models.evaluations.get
Make batch predictions	projects.locations.models.batchPredict	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get
Make online predictions	projects.locations.models.deploy	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
	projects.locations.models.predict	projects.locations.endpoints.predict

Back to top

AI Platform

Click one of the following user journeys:

• AI Platform: Train and deploy an XGBoost model with hosted runtime versions

• AI Platform: Train and deploy a scikit-learn model with hosted runtime versions

• AI Platform: Train and deploy a TensorFlow model with custom containers

• AI Platform: Train and deploy a TensorFlow model with hosted runtime versions

• AI Platform Prediction: Submit a batch prediction job for a hosted TensorFlow model

• AI Platform Training: Submit a hyperparameter tuning training job with TensorFlow

AI Platform: Train and deploy an XGBoost model with hosted runtime versions

To see an example of how to use Vertex AI to train and deply an XGBoost model with a prebuilt container, run the "Vertex AI Migration: Custom XGBoost model with prebuilt training container" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the AI Platform and Vertex AI products, and then use the following table to help migrate your API.

Step	AI Platform	Vertex AI
Train a model	projects.jobs.create	projects.locations.customJobs.create
	projects.jobs.get	projects.locations.customJobs.get
Deploy the model	projects.models.create	projects.locations.models.upload
	projects.models.versions.create
Make batch predictions	AI Platform batch prediction is not supported for XGBoost.	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get
Make online predictions	projects.predict	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
		projects.locations.endpoints.predict

Back to top

AI Platform: Train and deploy a scikit-learn model with hosted runtime versions

To see an example of how to use Vertex AI to train and deploy a scikit-learn model with a prebuilt container, run the "Vertex AI Migration: Custom Scikit-Learn model with prebuilt training container" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the AI Platform and Vertex AI products, and then use the following table to help migrate your API.

Step	AI Platform	Vertex AI
Train a model	projects.jobs.create	projects.locations.customJobs.create
	projects.jobs.get	projects.locations.customJobs.get
Deploy the model	projects.models.create	projects.locations.models.upload
	projects.models.versions.create
Make batch predictions	AI Platform batch prediction is not supported for scikit-learn.	projects.locations.batchPredictionJobs.create
		projects.locations.batchPredictionJobs.get
Make online predictions	projects.predict	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
		projects.locations.endpoints.predict

Back to top

AI Platform: Train and deploy a TensorFlow model with custom containers

To see an example of how to use Vertex AI to train and deploy a TensorFlow model with a custom container, run the "Vertex AI Migration: Custom Image Classification with custom training container" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the AI Platform and Vertex AI products, and then use the following table to help migrate your API.

Step	AI Platform	Vertex AI
Train a model	projects.jobs.create	projects.locations.customJobs.create
	projects.jobs.get	projects.locations.customJobs.get
Deploy the model	projects.models.create	projects.locations.models.upload
	projects.models.versions.create
Make batch predictions	projects.jobs.create	projects.locations.batchPredictionJobs.create
	projects.jobs.get	projects.locations.batchPredictionJobs.get
Make online predictions	projects.predict	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
		projects.locations.endpoints.predict

Back to top

AI Platform: Train and deploy a TensorFlow model with hosted runtime versions

To see an example of how to use Vertex AI to train and deploy a TensorFlow model with a prebuilt container, run the "Vertex AI Migration: Custom Image Classification with prebuilt training container" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the AI Platform and Vertex AI products, and then use the following table to help migrate your API.

Step	AI Platform	Vertex AI
Train a model	projects.jobs.create	projects.locations.customJobs.create
	projects.jobs.get	projects.locations.customJobs.get
Deploy the model	projects.models.create	projects.locations.models.upload
	projects.models.versions.create
Make batch predictions	projects.jobs.create	projects.locations.batchPredictionJobs.create
	projects.jobs.get	projects.locations.batchPredictionJobs.get
Make online predictions	projects.predict	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
		projects.locations.endpoints.predict

Back to top

AI Platform Prediction: Submit a batch prediction job for a hosted TensorFlow model

To see an example of how to use Vertex AI to submit a batch prediction job for a hosted TensorFlow model, run the "Vertex AI Migration: Custom Image Classification with prebuilt training container" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the AI Platform Prediction and Vertex AI products, and then use the following table to help migrate your API.

Step	AI Platform Prediction	Vertex AI
Train a model	projects.jobs.create	projects.locations.customJobs.create
	projects.jobs.get	projects.locations.customJobs.get
Deploy the model	projects.models.create	projects.locations.models.upload
	projects.models.versions.create
Make batch predictions	projects.jobs.create	projects.locations.batchPredictionJobs.create
	projects.jobs.get	projects.locations.batchPredictionJobs.get
Make online predictions	projects.predict	projects.locations.endpoints.create
		projects.locations.endpoints.deployModel
		projects.locations.endpoints.predict

Back to top

AI Platform Training: Submit a hyperparameter tuning training job with TensorFlow

To see an example of how to use Vertex AI to submit a hyperparameter tuning training job with TensorFlow, run the "Vertex AI Migration: Hyperparameter Tuning" Jupyter notebook in one of the following environments:

Open in Colab | Open in Colab Enterprise | Open in Vertex AI Workbench user-managed notebooks | View on GitHub

Read about the differences between the AI Platform Training and Vertex AI products, and then use the following table to help migrate your API.

Step	AI Platform Training	Vertex AI
Train a model	projects.jobs.create	projects.locations.hyperparameterTuningJobs.create
	projects.jobs.get	projects.locations.hyperparameterTuningJobs.get

Back to top

What's next

• Set up a project and a development environment to start using Vertex AI.