Yandex Compute Cloud release notes

Written by

Updated at February 13, 2026

Q3–Q4 2025

New VM platforms are now available:
- AMD Zen 4 (standard-v4a).
- AMD Zen 4 Compute-Optimized (highfreq-v4a).
- GPU PLATFORM V4 (gpu-standard-v4) with GPU cluster support.
Introduced the following new VM hardware generations on PCI_TOPOLOGY_V2 with increased disk and network interface limits:
- Gen 1.2 with a BIOS bootloader.
- Gen 2 with a UEFI bootloader.
Changed the logic of instance groups during a zonal incident to reduce impact and improve manageability.
Added support for zonal shift of instance groups.
Updated the logic of reserved instance pools; pool management is now also available in the management console.
Added support for generating SSH keys in the management console.
Introduced the yc compute maintenance command group to the Yandex Cloud CLI to manage VM maintenance.
Added the following Audit Trails management events:
- instancegroup.DisableZones: Zonal shift.
- instancegroup.EnableZones: Enabling availability zones.
Added support for the following new Yandex Monitoring metrics:
- filestore.index_cumulative_time: Total time of indexing operations in the file storage.
- filestore.index_latency: Delay when performing indexing operations in the file storage.
Updated the instance group deletion behavior: you can no longer pause processes in the group being deleted. Also, you cannot run any operations for paused instance groups.
Removed the simple VM creation form from the management console.

The reserved instance pool functionality is now on at the Preview stage. You can access it upon request.
Implemented a simple VM creation form in the management console.
Added sending the MigrateInstance data event to Yandex Audit Trails when dynamically migrating a VM with a maintenance policy configured.

The AMD Zen 3 platform (amd-v1) is launched at the Preview stage. You can access it upon request.
Upgraded monitoring and logging (observability) tools and improved the reliability of the high-speed secure Infiniband network in GPU clusters.
Added VM configuration with one GPU for the gpu-standard-v3i platform.

Released the disk encryption feature for General Availability.
Added VM configuration with two and four GPUs for the gpu-standard-v3i platform.
Added support for grouping gpu-standard-v3i VMs into a GPU cluster.
gpu-standard-v3-t4i is now available in the Kazakhstan region.

Implemented a new PCI topology. You can now create a VM with the new topology by running the yc instance create command with the --hardware-generation-id=generation2 and --hardware-features pci_topology=v2 parameters.
QEMU updated to version 7.

Now you can add and delete network interfaces without recreating VMs.
Added support for VHD.
Fixed a Terraform provider bug that occurred when creating a VM with multiple disks using the Count meta argument.
Added the description of gpus resources to the provider documentation.