You can put all your node pool to 0 VM but be careful to data lost (according to your node pool configuration, if you delete all the VM you can loose data). However, you will continue to pay for the control plane.

Another approach is to backup your data and to use IaC (Infra as code, such as terraform) to detroy and rebuild your cluster as needed.

Both approach are valid, they depend on your use case and how long to you need to hibernate your cluster.

An alternative is to use GKE Autopilot if your workloads are compliant with this deployment mode.