Cloud Functions is a serverless, event-driven compute platform designed specifically for the scenario described. It natively integrates with Google Cloud services like Cloud Storage, allowing it to be triggered automatically when a new image is uploaded. As a Function-as-a-Service (FaaS) offering, it excels at scaling from zero to handle sporadic spikes in traffic and automatically scales back down, ensuring you only pay for the compute time you use. This model perfectly aligns with the requirements to minimize both operational overhead (no servers to manage) and cost for idle time. Cloud Functions (2nd gen) is built on Cloud Run, providing enhanced performance and features suitable for production workloads.

A. Cloud Run is a strong serverless option but is better suited for containerized web applications or services. While it can be triggered by events via Eventarc, Cloud Functions offers a more direct, purpose-built, and simpler developer experience for this specific event-driven use case.

B. Compute Engine with an Auto-scaling Managed Instance Group is an Infrastructure-as-a-Service (IaaS) solution. It is not serverless and carries significant operational overhead for managing virtual machine instances, operating systems, and patching, contradicting the core requirement.

D. Google Kubernetes Engine (GKE) is a container orchestration platform. It introduces substantial operational complexity and cost, even with autoscaling. It is not a serverless compute service in its standard form and is overkill for a simple, event-driven function.

1. Google Cloud Documentation, "Choosing a serverless option": This document explicitly compares Cloud Functions and Cloud Run. It states, "For event-driven functions, Cloud Functions is a simple choice." It highlights that Cloud Functions is ideal for code that responds to events, such as a file upload to Cloud Storage.

Source: Google Cloud, Serverless compute options, "Cloud Functions".

2. Google Cloud Documentation, "Cloud Functions triggers": This page details the event-driven nature of Cloud Functions. It lists Cloud Storage as a direct event provider, stating, "You can trigger a function in response to changes in a Cloud Storage bucket," which is the exact use case in the question.

Source: Google Cloud, Cloud Functions Documentation, "Cloud Storage triggers".

3. Google Cloud Documentation, "Cloud Functions Overview": The overview emphasizes the "no-ops" and "pay-as-you-go" nature of the service. It states, "Cloud Functions automatically scales your application... Your function can scale from zero to thousands of instances without you having to do anything," directly addressing the scaling and cost requirements.

Source: Google Cloud, Cloud Functions Documentation, "Overview".