Releasesby GlyphIn version 1.36.2-gke.1498000 and later, GKE supports mixed-protocol Services of type LoadBalancer in general...
In version 1.36.2-gke.1498000 and later, GKE supports mixed-protocol Services of type LoadBalancer in general availability (GA). Mixed-protocol Services let both external (NetLB) and internal (ILB) passthrough Network Load Balancers handle simultaneous TCP and UDP traffic on a single IP address across IPv4, IPv6, and dual-stack environments. Main
Also shipped Jul 27, 2026Google Cloud in July 2026 →
AI_EXTRACT and AI_PARSE_DOCUMENT support for client-side encrypted stages and network-restricted accounts (Preview)
With this preview release, AI_EXTRACT and AI_PARSE_DOCUMENT support documents stored on stages that use client-side or server-side encryption, including in accounts that use PrivateLink or other network policies that restrict public network access to stages. For more information, see the following topics:
Custom incremental dynamic tables (General availability)
Custom incremental dynamic tables are now generally available. For more information, see Custom incrementalization .
[Launched] Generally Available: Claude Opus 5 on Azure Databricks
Azure Databricks now supports Anthropic Claude Opus 5 through AI Model Serving. Claude Opus 5 is Anthropic's latest model for advanced reasoning, coding, agentic workflows, and professional knowledge work. Claude Opus 5 is designed for complex tasks requi
GLM 4.7 reasoning_logprobs default will not ship before deprecation zai-glm-4.7 is scheduled for deprecation...
GLM 4.7 reasoning_logprobs default will not ship before deprecation zai-glm-4.7 is scheduled for deprecation on August 17, 2026 and will keep its existing default logprobs response shape until then. Requests that explicitly set the X-Cerebras-Version-Patch: 2 header still receive the separated reasoning_logprobs field.
Image Generation Models on OpenRouter
OpenRouter has introduced a dedicated /api/v1/images endpoint for image generation models, using the same billing and keys as chat completions.
Administration
AI credit controls for embed entity groups are now generally available. AI Fixed an issue where the Workload Identity Federation (WIF) role prefix wasn't properly configured across all AI-calling services, which could cause authentication failures when using Google Gemini Enterprise Platform (Vertex) as a custom AI model provider.
More from Google Kubernetes EngineEvery Google Kubernetes Engine release →
In GKE version 1.36 and later, GKE Dataplane V2 with NetworkPolicies supports up to 15,000 nodes per cluster...
In GKE version 1.36 and later, GKE Dataplane V2 with NetworkPolicies supports up to 15,000 nodes per cluster, increased from the previous limit of 7,500 nodes. For clusters exceeding 5,000 nodes, contact Cloud Customer Care to request a quota increase. For more information, see Cluster size limits and requirements .
GKE now supports opting out of the default kubernetes.io/arch=arm64:NoSchedule taint on Arm nodes in Standard...
GKE now supports opting out of the default kubernetes.io/arch=arm64:NoSchedule taint on Arm nodes in Standard node pools and in custom ComputeClasses. To opt out of the default taint, set the --node-architecture-taint-behavior gcloud CLI flag to NONE for a node pool or set the taintConfig.architectureTaintBehavior field to NONE for a ComputeClass.
GKE Gateway and Inference Gateway now support Cross-Origin Resource Sharing (CORS)
GKE Gateway and Inference Gateway now support Cross-Origin Resource Sharing (CORS). You can configure a CORS filter directly on an HTTPRoute resource by using the portable syntax standardized by Gateway API .
(2026-R32) Version updates
GKE cluster versions have been updated. New versions available for upgrades and new clusters. The following versions are now available for new GKE clusters, and for manual control plane upgrades and node upgrades for existing clusters. For more information about versioning and upgrades, see GKE versioning and support and About GKE cluster upgrades .
Note : Your clusters might not have these versions available
Note : Your clusters might not have these versions available. Rollouts are already in progress when we publish the release notes, and can take multiple days to complete across all Google Cloud zones. Version 1.36.2-gke.2064000 is now the default version for cluster creation in the Rapid channel.
(2026-R31) Version updates
GKE cluster versions have been updated. New versions available for upgrades and new clusters. The following versions are now available for new GKE clusters, and for manual control plane upgrades and node upgrades for existing clusters. For more information about versioning and upgrades, see GKE versioning and support and About GKE cluster upgrades .
Note : Your clusters might not have these versions available
Note : Your clusters might not have these versions available. Rollouts are already in progress when we publish the release notes, and can take multiple days to complete across all Google Cloud zones. Version 1.36.2-gke.1498000 is now the default version for cluster creation in the Rapid channel.
TPU Subslicing (also known as Dynamic Subslicing) is now generally available for Ironwood (TPU7x)
TPU Subslicing (also known as Dynamic Subslicing) is now generally available for Ironwood (TPU7x). This feature enables you to incrementally provision node pools for a cube or litepod, breaking them into smaller slices (subslices) to run workloads requiring smaller topologies.