You work for a social media company. You want to create a no-code image classification model for an iOS mobile application to identify fashion accessories. You have a labeled dataset in Cloud…

Serving predictions from a Vertex AI endpoint in the cloud introduces network latency, which does not align with the requirement for the 'lowest possible latency' for a mobile application.

Training with AutoML Edge is ideal for optimizing models for on-device deployment, offering a no-code solution. Exporting the model as a Core ML model ensures native integration and optimal performance on iOS devices, allowing predictions to run directly on the mobile app with the lowest possible latency by eliminating network round trips.

While AutoML Edge can export TFLite models for on-device deployment, Core ML is the native and often more optimized format for iOS applications, which is crucial for achieving the absolute lowest latency on Apple devices.

Similar to option A, using a Vertex AI endpoint for serving predictions from the cloud introduces network latency, failing to meet the lowest latency requirement for an iOS mobile application.