Compute quota tiers
Compute has two quota tiers: default and full. When you sign up, your Compute quotas are set to default values. Later, your quotas may be increased to full quota values automatically based on your usage history. Nebius determines whether and when to apply the full quota tier, and doesn’t send a notification when your quota tier changes. You can request increases to individual quotas, but you can’t control automatic changes to your quota tier. In the quota tables on this page, the Tier column shows a separate row for each tier when the default and full quotas differ. All means the same quota value applies to all tiers.Virtual machines
A VM and its resources count toward the quotas throughout the VM’s lifecycle, from its creation to deletion, regardless of whether it is running or stopped. These quotas apply to Compute VMs and also to Managed Service for Soperator and Managed Service for Kubernetes® nodes, which are based on Compute VMs.All virtual machines
GPU virtual machines
These quotas limit the number of NVIDIA® GPUs attached to regular VMs. GPUs allocated from reservations don’t count towards them.Non-GPU virtual machines
Storage
InfiniBand
GPU cluster quotas apply to Compute GPU clusters and GPU clusters used by Managed Soperator or Managed Kubernetes.Network
See Quotas in Virtual Networks.Quota usage recommendations
Best practices for using VM quotas:- When you create new VMs, calculate your available quotas, even if you plan to keep the VMs mostly in a stopped state.
- From time to time, review and delete unused VMs to free up quotas.
- If you need to temporarily free up VM resources, delete VMs instead of just stopping them.
InfiniBand and InfiniBand Trade Association are registered trademarks of the InfiniBand Trade Association.