Google today enhanced its Google Cloud Platform (GCP) public cloud infrastructure offering with new virtual machine (VM) instances that boast up to 64 virtual central processing unit (CPU) cores, effectively providing more computing power for applications. They’re available now in beta for all users and projects out of Google’s data center zones that run Intel’s Haswell and Broadwell chips.

There are three available flavors in the Google Compute Engine (GCE): a standard instance called n1-standard-64, a high-memory instance called n1-highmem-64, and an instance with high CPU-to-memory ratio called n1-highcpu-64. They’re all available from Google’s web-based Cloud Console and in the command-line interface (CLI) with the Google Cloud SDK.

Google has been taking many steps to become more competitive with other public clouds. The biggest one, Amazon Web Services (AWS), does offer machines with 64 vCPUs. There are even 128 vCPU machines available from AWS. Currently, Microsoft Azure, the second biggest cloud, does not have machines with more than 32 vCPUs.

Google followed AWS and Azure and introduced a key management service in January.

People with free trials will need to ask for a quota increase in order to spin up one of the new instances — if you try to create one with a free trial, you’ll receive an error saying you’ve reached a limit of 24 CPUs.

“We’ll improve the way quotas work in the near future so that it will be possible to create a 32-core instance for new, free trial users, a Google spokesperson told VentureBeat in an email.

Google’s prowess in getting big customers for its cloud was proven last month, when Snap, the newly public company behind Snapchat, revealed that it had a $2 billion contract with the Google Cloud.