21 releases · 1 this month
- Added: Managed Service for Kubernetes® now supports dynamic resource allocation (DRA) on clusters running Kubernetes 1.34 or later. After you enable DRA on a node group , workloads can request NVIDIA GPUs through the NVIDIA DRA driver, which you need to install and maintain, and RDMA or InfiniBand™ network interfaces through DRANet, which Nebius manages.
Nebius has made spot prices for GPU platforms public and added the ability to restart Serverless AI endpoints.
- Added: Cloud Shell is now available in preview. It's a browser-based terminal in the web console with the Nebius AI Cloud CLI, Terraform, Git and Python preinstalled, and the CLI is already authenticated with your user account. Sessions last up to one hour.
- Added: Compute documentation now includes how to change the preset of an existing VM to give it more or fewer vCPUs and RAM. You need to stop the VM to change its preset. Added: You can now use several shared filesystems with one CSI driver installation in Managed Service for Kubernetes®. Each filesystem gets its own storage class.
- Added: You can now connect Wiz to Nebius AI Cloud to scan the configuration of your Compute, Managed Service for Kubernetes®, Virtual Networks and Object Storage resources for misconfigurations.
- Added: Spot pricing now applies to preemptible VMs and services that run on them. Nebius sets a current spot price for each preemptible GPU platform, which can change every 15 minutes. When you create a preemptible resource, select a pricing policy to cap the maximum price, or choose Follow spot price .
- Added: DNS in Virtual Networks is now available. Create DNS zones attached to a network, add records that resolve only from workloads in that network and give your VMs stable internal endpoints under a domain name of your choosing. Added: The new eu-south1 (Spain, Madrid) private region is now available.
- Added: Filesystem buckets in Object Storage are now available in preview. Attach an Object Storage bucket to an existing shared filesystem to read or write the same data over both an S3-compatible API and the filesystem interface. Filesystem buckets are excluded from the Object Storage SLA while in preview.
- Added: New guide: Authenticate the Nebius AI Cloud CLI in environments without a browser , such as Docker containers, CI/CD runners and dev containers. The guide covers service account authentication with an authorized key and user account authentication by opening an authorization link on another machine.
- Added: You can now request plain HTTP access to the Object Storage S3 API for selected buckets and use http:// instead of https:// in the S3 API endpoint. Choose between the Region-local mode (access only from Nebius AI Cloud VMs in the bucket's region) and the All mode (access from any source).
- Added: The new us-north1 (Woodbury, Minnesota, US) private region is now available. It supports Compute with the NVIDIA® B300 NVLink and Non-GPU AMD EPYC Genoa platforms , Object Storage , Managed Service for Kubernetes® , Container Registry , Key Management Service and other Nebius AI Cloud services.
- Added: New tutorials for Nebius Tunnels : deploy a large language model on a GPU VM and run JupyterLab on a VM , reaching each service from the internet through a tunnel while the VM itself has no public IP address. Added: Managed Service for Kubernetes® CSI driver for shared filesystems is now available as chart version 0.1.7 .
- Added: The Nebius AI Cloud REST API is now available. It exposes the same resources and operations as the gRPC API over HTTP/1.1 with JSON. See the quickstart , authentication and API reference for details.
- Added: You can now run Serverless Containers jobs and endpoints on preemptible VMs . Pass --preemptible when you create a job or endpoint to use a cheaper GPU VM that Compute can stop at any time. Added: Object Storage bucket policy rules now support the ? placeholder in paths . Each ?
- Added: You can now set project limits (capacity allowance) on a capacity block group to cap how many GPUs each project in your tenant can consume. See Managing project limits for how to view, set, remove and disable limits from the web console or CLI.
- Added: The new eu-west2 (Paris, France) private region is now available. It supports Compute with the NVIDIA® B300 NVLink and Non-GPU AMD EPYC Genoa platforms , Object Storage , Managed Service for Kubernetes® , Managed Service for Soperator , Container Registry and other Nebius AI Cloud services. Added: The serial console is now available in preview.
- Added: You can now create disk snapshots in Compute to capture a point-in-time copy of a disk. Use snapshots to back up disks before risky operations, clone disk state or restore data from a snapshot. Added: Managed Service for Kubernetes® node groups now integrate with Virtual Networks security groups .
- Added: Managed Service for Soperator now supports ephemeral nodes . You can release specific worker nodes from a Soperator cluster and reuse the same capacity block group for inference workloads in a separate Managed Service for Kubernetes® cluster, without taking down the entire Soperator cluster.
- Added: Nebius Tunnels is now available in preview. You can expose local HTTP and TCP services to the internet through an outbound agent connection, without opening inbound firewall ports.
- Added: The Key Management Service is now available to Nebius AI Cloud users. It enables you to issue and store cryptographic keys, so you can protect and secure your applications. Added: Clusters and node groups in Managed Service for Kubernetes® now support Kubernetes 1.34 . It is now the default and recommended Kubernetes version in Nebius AI Cloud .
- Added: A free trial is now available for individual users in the eu-north1 region. The trial includes 30 days of access and $5 in credit to explore the platform. Added: You can now use the Nebius AI Cloud SDK for Go to interact with Nebius AI Cloud services from Go applications.
Releasesby Glyph