In GKE Autopilot mode, what changes compared with Standard mode?

Why this is the answer

In Autopilot, Google provisions, scales and maintains the nodes, and billing follows the CPU, memory and ephemeral storage each Pod requests rather than any VM shape. The second option describes Standard mode, where Google runs only the control plane and you pay for the node VMs. Google never manages your workloads themselves, and no cluster-uptime-only billing model exists. Network charges are real but are not what separates the two modes.