A data scientist has a new model that outperforms the current production model on test data. They want to run A/B testing in the production SageMaker endpoint to verify performance on real traffic before fully replacing the model. Which steps are required to perform A/B testing? (Choose two.)

To perform A/B testing in SageMaker, you need to define multiple model variants within a single endpoint. The first correct step is to create a new endpoint configuration that specifies a production variant for each model you want to test (the existing model and the new model). This configuration allows you to assign traffic weights to each variant. The second correct step is to update the existing endpoint to use this new endpoint configuration. This action applies the new configuration to the live endpoint, enabling traffic splitting between the models. Incorrect options: "Create a new endpoint configuration that contains two target variants that point to different endpoints" is incorrect because A/B testing involves multiple models within one endpoint, not across different endpoints. "Deploy the new model to the existing endpoint" is insufficient; you need to configure how traffic is routed, which requires an endpoint configuration update. "Update the existing endpoint to activate the new model" is too vague and doesn't specify the necessary configuration changes for A/B testing.