It seems like you have been trying to create a regular virtual machine instance with the 'Preemptible GPU quota' on your project. You might need to create a 'Preemptible VM Instance' to use your 'preemptible GPU quota' . Check this Google Cloud Platform blog for more details about ‘Preemptible GPUs’ and its usage.
That's being said, you can always request and add a regular GPU on any virtual machine instance. But, check the restrictions mentioned in the documentation before operating the GPU integrated instances.