Google Cloud release notes

You can now use continuous queries to export BigQuery data to Spanner in real time. This feature is in Preview.

Conversational Agents: Routine playbooks are now generally available.

Conversational Agents: Parameter passing is now available and documented between routine playbooks, task playbooks, and flows.

Use Gemini Code Assist agent mode in IntelliJ (Preview)

Stay in control with interactive reviews and approvals

You can use Gemini Code Assist chat in agent mode to complete complex, multi-step tasks and goals with complete control over every change. Before making any modifications, the agent will present a plan for your review. You can edit, ask for changes, approve, or deny any suggested changes. This collaborative approach combines the power of AI with your expertise, resulting in better code and a more efficient workflow.

To get started with agent mode, restart your IDE and follow the instructions in Use agentic chat as a pair programmer.

Auto Approve mode lets the agent act on your behalf (Preview)

Enable auto approve mode to let the agent act on your behalf. Once the agent is done you can review and roll back changes as you see fit.

Use agent mode in Gemini Code Assist for VS Code (Preview)

Tackle complex tasks with Gemini Code Assist agent mode for VS Code

Gemini Code Assist agent mode is available for all users. Describe your goal, and Gemini will create a plan for you to review and approve before any code is changed. This update removes the insiders channel requirement, and includes persistent agent mode or interactive chat state between IDE restarts, real-time shell command output, and faster UI performance. This collaborative approach combines AI power with your expertise, enabling you to complete multi-step tasks with complete control and efficiency.

To get started with agent mode, see Use agentic chat as a pair programmer.

Effectively collaborate with your agent with the improved diff view functionality (Preview)

We've enhanced Gemini Code Assist agent mode with powerful new editing capabilities. You can edit code changes directly in the integrated Diff view for precise, on-the-fly adjustments. To improve clarity, we've also added inline diffs directly in the chat, making it easier to see proposed changes at a glance. These features give you unparalleled control and a more efficient way to refine AI-suggested code.

Support for Node.js 24 runtime is in Preview. Node.js 24 is in the Current release state and enters long-term support (LTS) in October 2025. For more information, see Node.js v24.0.0 (Current) in the Node.js website.

Support for Go 1.24 runtime is in General Availability (GA).

Support for Go 1.25 runtime is in Preview. This runtime is available for early testers using existing release candidates.

Support for Node.js 24 runtime is in Preview. Node.js 24 is in the Current release state and enters long-term support (LTS) in October 2025. For more information, see Node.js v24.0.0 (Current) in the Node.js website.

The following Vertex AI supported resources are now generally available (GA):

• Dataset items
• Featurestore containers
• MetadataStore instances
• Model resources

Database server compatibility with PostgreSQL version 17 is now available in Preview. You can create AlloyDB clusters with PostgreSQL 17 compatibility.

Server-sent events and EventFlows are Generally Available (GA)

Apigee supports continuous response streaming from server-sent event (SSE) endpoints to clients in real time. The Apigee SSE feature is useful for handling large language model (LLM) APIs that operate most effectively by streaming their responses back to the client. SSE streaming reduces latency, and clients can receive response data as soon as it is generated by an LLM. This feature supports the use of AI agents that operate in real time environments, such as customer service bots or workflow orchestrators. For more information, see Streaming server-sent events.

Streaming from SSE endpoints is available in Apigee and in Apigee hybrid v1.15.0 and newer.

You can now associate data policies directly on columns. This feature enables direct database administration for controlling access and applying masking and transformation rules at the column level. This feature is in Preview.

Global external Application Load Balancers now support the JA4 fingerprint. The JA4 fingerprint can be added to a custom request header using the tls_ja4_fingerprint variable.

This capability is now in General Availability.

Inline diff (Preview)

VS Code Gemini Code Assist 2.42.0

Inline diff is available for VS Code Gemini Code Assist 2.42.0 in Preview. With the inline diff view, you can visually highlight code changes directly in your editor—green for additions, red for deletions. This feature boosts productivity by offering immediate visibility on modifications, streamlining reviews, and enabling precise control over integrating new code, all without leaving your current view.

If preferred, you can change the diff view settings to have a side-by-side diff window instead of the inline diff.

(New guide) Well-Architected Framework: Financial services industry (FSI) perspective: Principles and recommendations that are specific to FSI, aligned to each pillar of the Architecture Framework.

In GKE version 1.33.2-gke.1335000 and later, the GKE Gateway controller supports Gateway API v1.3 CRDs.

In GKE version 1.33.1-gke.1788000 and later, you can target specific reservation sub-blocks in a reservation block by using the reservationSubBlock field in compute classes.

In GKE version 1.32.2-gke.1359000 and later, you can now configure collection scheduling for single-host and multi-host TPU node pools by using compute classes. Collection scheduling lets you set a Service Level Objective (SLO) for your TPU workloads.

In GKE version 1.33.2-gke.1335000 and later, the GKE Gateway controller supports Gateway API v1.3 CRDs.

In GKE version 1.33.1-gke.1788000 and later, you can target specific reservation sub-blocks in a reservation block by using the reservationSubBlock field in compute classes.

In GKE version 1.32.2-gke.1359000 and later, you can now configure collection scheduling for single-host and multi-host TPU node pools by using compute classes. Collection scheduling lets you set a Service Level Objective (SLO) for your TPU workloads.

Automate tasks with Playbook Loops

This feature is in Preview.

Playbook functionality has been enhanced to include Playbook Loops. This feature update lets playbooks iterate over lists or entities, performing one or more actions for each item. It streamlines automation by eliminating the need for duplicated steps or custom actions when processing multiple items. You can configure Playbook Loops directly within a playbook or inside a playbook block.

For setup instructions and use case examples , see Automate tasks with Playbook Loops.

Playbook Simulator enhancements for loops

The Playbook Simulator now supports visualization and debugging of playbooks that contain loops. This lets you clearly see and navigate through each loop iteration within the simulator viewer.

Additionally, the step display order has been updated to show actions from top to bottom (oldest at the top, newest at the bottom), with automatic scrolling to the most recent activity.

For more details, see Loops in the Playbook Simulator.

Automate tasks with Playbook Loops

This feature is in Preview.

Playbook functionality has been enhanced to include Playbook Loops. This feature update lets playbooks iterate over lists or entities, performing one or more actions for each item. It streamlines automation by eliminating the need for duplicated steps or custom actions when processing multiple items. You can configure Playbook Loops directly within a playbook or inside a playbook block.

For setup instructions and use case examples , see Automate tasks with Playbook Loops.

Playbook Simulator enhancements for loops

The Playbook Simulator now supports visualization and debugging of playbooks that contain loops. This lets you clearly see and navigate through each loop iteration within the simulator viewer.

Additionally, the step display order has been updated to show actions from top to bottom (oldest at the top, newest at the bottom), with automatic scrolling to the most recent activity.

For more details, see Loops in the Playbook Simulator.

You can now attach tags to Cloud Data Fusion instances during instance creation. This ensures immediate metadata for better organization, cost tracking, and policy automation. For more information, see Attach tags during instance creation.

You can create a Remote Direct Memory Access (RDMA) over converged ethernet (RoCE) Virtual Private Cloud (VPC) network and configure firewall rules that apply to the network. For more information, see Cloud NGFW for RoCE VPC networks. This feature is available in Preview.

Advanced load balancing for managed Cloud Service Mesh (TD) now generally available (GA).

Google Cloud NetApp Volumes now supports cross-project cross-region replication for Standard, Premium, and Extreme service levels. This feature is generally available for allow-listed users. For more information, see About volume replication.

Google Cloud NetApp Volumes now supports volume backups for large capacity volumes. This feature is now generally available. For more information, see About backups.

Web server restarting is is now generally available (GA) in Cloud Composer 2 and Cloud Composer 3.

You can now monitor and understand the costs and utilization of resources in your Google Cloud project or App Hub application by using the Cost Explorer. This feature is in Public Preview. For more information, see Optimize costs with the Cost Explorer.

Generally available: The general-purpose C4 machine series now supports Hyperdisk Balanced High Availability. For more information, see Supported disk types for C4.

Conversational Agents: 21 new Chirp 3 HD voices are now available across 34 locales.

Conversational Agents data stores: Conversational Agents now supports AlloyDB AI, Bigtable, Firestore, Spanner and Cloud SQL data store source as public GA features; Microsoft Entra ID source has now private GA support. See the data store documentation for a complete list.

Stop in-progress chat responses with IntelliJ

IntelliJ Gemini Code Assist 1.22.1

You can stop chat responses with IntelliJ Gemini Code Assist 1.22.1. Undesired long running or errant chat responses are immediately halted.

The Oracle JDBC driver has been updated to version 19.25.

For faster response time for queries in BigQuery, Looker will execute BigQuery queries by using jobCreationMode=JOB_CREATION_OPTIONAL. If BigQuery can return immediate results, it will run the query without creating a job, so the record in the Looker query history will have a BigQuery query ID instead of a BigQuery job ID. See the Understanding query performance metrics documentation page for more information about the BigQuery BI Engine metrics.

The Query Concurrency System Activity Explore is now available. This Explore can help you identify periods of high load and investigate performance bottlenecks that are related to database connection limits.

New report canvas sizes

Two new preset canvas size options are available, letting you control the width and height of your report on the screen. The new options are size A4, available in portrait or landscape orientations.

Learn more about report and page layout options.

New alt text field for images

You can now add alt text to report images to make them accessible to screen readers. See the Web Content Accessibility Guidelines (WCAG) guidelines on alt text for more information about writing alt text.

Looker connector enhancements

Additional calculated field functions are now available in Preview.

Performance improvement for BigQuery data sources

Report viewers may notice improved performance when the report uses a BigQuery data source.

The BigQuery connector supports short query optimized mode. In this mode, when BigQuery determines it can finish a query quickly, BigQuery prioritizes returning immediate results instead of creating a BigQuery job. Short query optimized mode may apply in the following situations:

• When the data source uses Viewer's Credentials
• When the data source uses Owner's Credentials but you are not the credential owner.

There is no change in user experience or report behavior for data source credential owners.

Highlight charts by filter

The Applied Filters panel now lets you click on a filter to highlight all charts that the filter applies to.

Add descriptions to reports

You can now add text descriptions to a report. The report search now matches the report title as well as the description.

For the Enterprise service tier, Security Command Center offers data residency support in the European Union, Saudi Arabia, and United States. This feature is in General Availability.

To upload your Wasm plugin code to Artifact Registry, you can use generic format repositories, in addition to Docker repositories. This feature is in Preview.

Preview stage support for the following integration:

• License Manager

Cloud Armor supports internal service security policies for the service mesh to enforce global server-side rate limiting per client in Preview.

Siemplify: Version 94.0

• The following new actions have been added:

  • Get Custom Field Values

  • Resume Case SLA

  • Pause Case SLA

You can now use the VECTOR_INDEX.STATISTICS function to calculate how much an indexed table's data has drifted between when a vector index was created and the present. If table data has changed enough to require a vector index rebuild, you can use the ALTER VECTOR INDEX REBUILD statement to rebuild the vector index. This feature is in Preview.

Access Transparency supports BigQuery data preparation in the GA stage.

The CREATE EXTERNAL TABLE and LOAD DATA statements now support the following options in Preview:

• null_markers: define the strings that represent NULL values in CSV files.
• source_column_match: specify how loaded columns are matched to the schema. You can match columns by position or by name.

You can now use the MATCH_RECOGNIZE clause in your SQL queries to filter and aggregate matches across rows in a table. This feature is in Preview.

Preview: Multi-writer support for Hyperdisk Extreme disks. You can give up to 16 instances simultaneous read-write access to the same disk. For more information, see Share disks between instances.

Custom extractor model pretrained-foundation-model-v1.5-pro-2025-06-20 powered by Gemini 2.5 Pro is in Public Preview. It has ML processing available for US and EU regions, and 30 page per minute processing requests.

For more information, see Managing processor versions.

(New guide) Secure apps and resources by using context-aware access: Describes how you can use context-aware access to secure different types of apps and resources.

(New guide) Best practices for securing apps and resources by using context-aware access: Describes best practices for using context-aware access to secure apps and resources.

Silent Host Monitoring

New configuration options are now available for Silent Host Monitoring. You can now define detection rule-based Silent Host Monitoring in SecOps using UDM fields or labels, configurable within a specified time window.

For more information, see Silent host monitoring.

Silent Host Monitoring

New configuration options are now available for Silent Host Monitoring. You can now define detection rule-based Silent Host Monitoring in SecOps using UDM fields or labels, configurable within a specified time window.

For more information, see Silent host monitoring.

Access Transparency supports BigQuery data preparation in the GA stage.

You can now use the DISTINCT pipe operator to select distinct rows from a table in your pipe syntax queries. This feature is generally available (GA).

Generally available: The general-purpose C4D machine series offers the following bare metal machine types:

• c4d-standard-384-metal
• c4d-highcpu-384-metal
• c4d-highmem-384-metal

This is the first machine series to offer AMD-based bare metal instances. Bare metal instances let you create an instance with direct access to the machine's CPU and memory, without a virtualization layer in the middle. C4D uses Titanium to deliver more compute and memory resources for your workloads by offloading network and I/O processing from the host hardware. To learn more, see C4D machine series. For information about bare metal instances, including regional availability, see Bare metal instances on Compute Engine.

Added ARM support for the Lustre 2.14.0 drivers.

Added ARM support for the Lustre 2.14.0 drivers.

In GKE version 1.33.2-gke.1111000 and later, you can use compute classes to set Kubernetes labels on all nodes that are created for that compute class. These labels are applied to the corresponding Node objects in the Kubernetes API. For more information about setting node labels in compute classes, see the ComputeClass custom resource definition.

In GKE version 1.33.2-gke.1111000 and later, you can use compute classes to set Kubernetes labels on all nodes that are created for that compute class. These labels are applied to the corresponding Node objects in the Kubernetes API. For more information about setting node labels in compute classes, see the ComputeClass custom resource definition.

New parser documentation now available

New parser documentation is available to help you ingest and normalize logs from the following sources:

Collect Apache Tomcat logs

Collect Appian Cloud logs

Collect Archer IRM logs

Collect ArcSight CEF logs

Collect Area 1 logs

Collect Aruba EdgeConnect SD-WAN logs

Collect Atlassian Cloud Admin Audit logs

Collect Avatier logs

Collect Avigilon Access Control Manager logs

Collect AWS CloudTrail logs

Collect Barracuda CloudGen Firewall logs

Collect Barracuda Web Filter logs

Collect Broadcom CA PAM logs

Collect Broadcom SSL VA logs

Collect Cato Networks logs

Collect Check Point Harmony logs

Collect CipherTrust Manager logs

Collect Cisco VCS logs

Collect Cisco VPN logs

Collect Cisco WSA logs

Collect CyberArk Privilege Cloud logs

Collect Digi Modems logs

Collect F5 DNS logs

Collect F5 VPN logs

Collect Forcepoint CASB logs

Collect HPE BladeSystem c7000 logs

Collect Skyhigh Security logs

Collect Trellix IPS logs

New parser documentation now available

New parser documentation is available to help you ingest and normalize logs from the following sources:

Collect Apache Tomcat logs

Collect Appian Cloud logs

Collect Archer IRM logs

Collect ArcSight CEF logs

Collect Area 1 logs

Collect Aruba EdgeConnect SD-WAN logs

Collect Atlassian Cloud Admin Audit logs

Collect Avatier logs

Collect Avigilon Access Control Manager logs

Collect AWS CloudTrail logs

Collect Barracuda CloudGen Firewall logs

Collect Barracuda Web Filter logs

Collect Broadcom CA PAM logs

Collect Broadcom SSL VA logs

Collect Cato Networks logs

Collect Check Point Harmony logs

Collect CipherTrust Manager logs

Collect Cisco VCS logs

Collect Cisco VPN logs

Collect Cisco WSA logs

Collect CyberArk Privilege Cloud logs

Collect Digi Modems logs

Collect F5 DNS logs

Collect F5 VPN logs

Collect Forcepoint CASB logs

Collect HPE BladeSystem c7000 logs

Collect Skyhigh Security logs

Collect Trellix IPS logs

You can ask Gemini for predefined role suggestions using the IAM role picker in the Google Cloud Console. This feature is in preview.

For more information, see Get predefined role suggestions with Gemini assistance.

You can set up the Agent for Compute Workloads to collect metrics of your Oracle workloads running on Google Compute Engine instances. For more information, see Set up the agent for Oracle.

Generally available: You can troubleshoot workloads with slow performance by using straggler detection metrics and logs.

Stragglers are single-point, non-crashing failures that eventually slow down your entire workload. Large-scale ML workloads are very susceptible to stragglers, and VMs with stragglers are often very difficult to notice and pinpoint without straggler detection.

For more information, see Monitor VMs and Slurm clusters and Troubleshoot slow performance.

Apigee and hybrid plugin instance management

You can now create and delete plugin instances for Apigee and Apigee Hybrid while associating the respective Apigee runtime projects to API hub.

For more information, see Auto-register Apigee proxies.

The following resource types are now publicly available through the ExportAssets, ListAssets, BatchGetAssetsHistory, QueryAssets, Feed, and Search (SearchAllResources, SearchAllIamPolicies) APIs.

• Service Management API
  • servicemanagement.googleapis.com/Service

Generally available: You can create instant and standard snapshots from Hyperdisk volumes in multi-writer mode. You can also clone Hyperdisk volumes in multi-writer mode.

For more information, see Share disks between instances.

Quality AI offers multiple scorecards in preview. With multiple scorecards, Quality AI can evaluate a single conversation against different criteria and provide multiple conversation scores. You can also choose which scorecard data to view on each page of the console.

The Data Boundary for Impact Level 4 (IL4) now supports the following products:

• Certificate Authority Service
• Cloud Build
• Cloud NAT
• GKE Hub
• Memorystore for Redis
• Secret Manager
• Sensitive Data Protection

The Data Boundary for Impact Level 5 (IL5) now supports the following products:

• Certificate Authority Service
• Cloud Build
• Cloud NAT
• Cloud VPN
• GKE Hub
• Memorystore for Redis
• Secret Manager
• Sensitive Data Protection

See Supported products by control package for more information.

You can now use the WITH pipe operator to define common table expressions in your pipe syntax queries. This feature is generally available (GA).

You can now use named windows in your pipe syntax queries. This feature is generally available (GA).

Highly resilient environments are now generally available (GA) in Cloud Composer 3.

Application-specific resource attributes are attached to your trace data when your App Hub applications use supported Google Cloud resources, or when you instrument an application with OpenTelemetry and use the Google Cloud Telemetry endpoint. To learn more, see the following:

• Application Monitoring overview
• View application telemetry
• Find and explore traces
• Telemetry (OTLP) API overview

Application-specific resource attributes are attached to your trace data when your App Hub applications use supported Google Cloud resources, or when you instrument an application with OpenTelemetry and use the Google Cloud Telemetry endpoint. You can use the Trace Explorer to filter by your application, your service, or your workload. To learn more, see the following:

• Find and explore traces
• Application Monitoring overview
• View application telemetry
• Telemetry (OTLP) API overview

Conversational Agents: CMEK is now available in EU regions.

Conversational Agents: The conversational history flow analysis feature is now available.

Conversational Agents: The model gemini-2.5-flash is now available in all regions. This change applies to the following features:

• Generators
• Playbooks (Public Preview)
• Data stores tools (Public Preview)

The Detailed Disbursements and Customer Incremental Insights ISV reports are now available in BigQuery. These reports leverage BigQuery's out-of-the-box connectors to aggregate your Google Cloud Marketplace data with other data sources for your analysis, reporting, and data visualization needs.

Learn how to enable these new reports from the Producer Portal.

Report abuse

Looker Studio report viewers can now report abusive content. Content is automatically reviewed, and content that is reported as abusive is hidden or deleted.

Performance improvement for BigQuery data sources

Report viewers may notice improved performance when the report uses a BigQuery data source.

The BigQuery connector supports short query optimized mode. In this mode, when BigQuery determines that it can finish a query quickly, BigQuery prioritizes returning immediate results instead of creating a BigQuery job. Short query optimized mode may apply in the following situations:

• When the data source uses Viewer's Credentials
• When the data source uses Owner's Credentials but you are not the credential owner

There is no change in user experience or report behavior for data source credential owners.

The following Container Threat Detection detectors for file monitoring are in Preview:

• Collection: Pam.d Modification
• Credential Access: Access Sensitive Files on Nodes
• Defense Evasion: Disable or modify Linux audit system
• Defense Evasion: Root Certificate Installed
• Execution: Suspicious Cron Modification
• Persistence: Modify ld.so.preload

You can now add comments to notebooks, data canvases, data preparation files, or saved queries. You can also reply to existing comments or get a link to them. This feature is in Preview.

You can now create BigQuery ML models by using the Google Cloud console user interface. This feature is in Preview.

Spend-based committed use discount (CUD) metadata export to BigQuery (public preview)

You can now access spend-based CUD metadata programmatically through a BigQuery export. This data provides a comprehensive, daily snapshot of spend-based CUDs, which you can join with other billing data exports for improved CUD reporting and management.

Learn more about the CUD metadata export.

You can disable the built-in run.app URL of a Cloud Run service to ensure that traffic can only ingress through paths that you've explicitly configured (GA).

Added Gemma 3 fine-tuning notebook using Axolotl docker with support for 1b, 4b, 12b, and 27b variants.

Google Chronicle: Version 62.0

• The following new actions have been added:

  • Remove Rows From Data Table

  • Get Data Tables

  • Is Value In Data Table

  • Add Rows To Data Table

You can now use Valkey GLIDE to connect your applications to Memorystore for Valkey instances.

Added Gemma 3 fine-tuning notebook using Axolotl docker with support for 1b, 4b, 12b, and 27b variants.

Support for Go 1.24 runtime is in Preview.

Support for Go 1.24 runtime is in Preview.

You can now commercialize your BigQuery sharing listings on Google Cloud Marketplace. This feature is generally available (GA).

You can flatten JSON columns in BigQuery data preparation with a single operation. This feature is generally available (GA).

Compute flexible committed use discounts (CUDs) have expanded to also cover your Cloud Billing account's spend across Cloud Run services with request-based billing and Cloud Run functions.

The improved spend-based CUD experience is available without requiring an opt-in for new users or users who don't have an active spend-based CUD.

To learn more about how opting into the new model affects your flexible CUDs, see Committed use discounts.

Support for the Go 1.24 runtime is in Preview.

Support for the Go 1.24 runtime is in Preview.

Dataproc now allows Dynamic update of multi-tenancy clusters.

You can now create an AlloyDB instance with a specific IP address range using the Google Cloud CLI, Terraform, or REST API. You can also override IP address range allocations configured during cluster creation. For more information, see Create an instance with a specific IP address range. This feature is generally available GA.

Support for editing and deleting security actions

With this release you can edit and delete existing security actions using either the UI or the Apigee Management APIs.

For usage information, see the security actions documentation.

Updates to the automatic cataloging of Dataform metadata in Dataplex improve the near real-time management and search capabilities for repository metadata. These features are generally available (GA).

Multimodal MedGemma 27B IT, MedSigLIP, and T5Gemma models are available through Model Garden.

Spanner Data Boost supports data stored on hard disk drives (HDD). This feature is generally available (GA).

Preview stage support for the following integration:

• Address Validation API

• Places (New) API

Multimodal MedGemma 27B IT, MedSigLIP, and T5Gemma models are available through Model Garden.

The following resource types are now publicly available through the ExportAssets, ListAssets, BatchGetAssetsHistory, QueryAssets, and Feed APIs.

• Parameter Manager
  • parametermanager.googleapis.com/Parameter
  • parametermanager.googleapis.com/ParameterVersion

Notebook Security Scanner is a built-in package vulnerability detection service of Security Command Center. This feature is available in Preview to the Security Command Center Premium or Enterprise tier.

You can enable and use Notebook Security Scanner to detect vulnerabilities in Python packages that are used in Colab Enterprise notebooks (files with the ipynb filename extension) and resolve those package vulnerability findings.

To reduce the cost of running your inference jobs, you can now use flex-start VMs, which are powered by Dynamic Workload Scheduler. Flex-start VMs offer significant discounts and are well-suited for short-duration workloads. This feature is available in Preview.

For more information, see Use DWS flex-start VMs with inference.

Generally available: You can now manage the Collective Communication Analyzer (CoMMA), a library that uses the NVIDIA Collective Communication Library (NCCL) profiler plugin to collect detailed NCCL telemetry for GPU machine types. The collected performance metrics and operational events are used for analyzing and optimizing large-scale AI and ML training workloads.

CoMMA is automatically installed and enabled on A4X, A4 High, and A3 Ultra machine types when using specific images. You can manage this data collection by disabling the plugin, adjusting its data granularity levels, or manually installing it on other GPU machine types. For more information, see Enable, disable, and configure CoMMA.

Restrict email transfers

You can now configure your instance to prevent users with the agent role from transferring email sessions to other agents. Agents can still assign unassigned emails to themselves, and users with the manager role can still transfer email sessions from agent to agent.

Administrators: There's a new Transfer Restrictions pane at Settings > Queue > Email > Edit / View > [queue] > Transfer Restrictions > Configure.

Skip the connecting message playback

You can now configure your instance to skip playback of the connecting message when calls are connected to agents.

Administrators: The Settings > Call > Call Details pane contains a new Skip the Connecting Message playback checkbox.

Workforce Management terminology update

We've updated the terminology in the Workforce Management interface to align with Google Cloud CCaaS terminology. For example, we've changed "supervisor" to "manager", "employee" to "agent", and "Supervisor Portal" to "Manager Portal".

Generative knowledge assist is available in Agent Desktop

Generative knowledge assist is now available in Agent Desktop as a widget that you can drag into a desktop panel.

For more information, see Create desktop panels.

Generative knowledge assist is available in the agent adapter

Generative knowledge assist is now available in the agent adapter.

Web SDK version 3.37

Starting with version 3.37, web SDK releases align with portal releases and share the same version number.

Web SDK version 3.37 includes the following update: we've improved the accessibily of the web SDK to be in compliance with the European Accessibility Act.

New alignment option for Cartesian charts

The new Align with grid setting lets you set the alignment of the chart legend with the position of the chart grid, instead of aligning with the chart title.

This setting is available for Cartesian charts in reports that have modern charts enabled.

M131 release

The M131 release of Vertex AI Workbench instances includes the following:

• Updated the Dataproc JupyterLab plugin to version 0.1.89.

Application Load Balancers and Proxy Network Load Balancers now support TLS certificates with large key sizes. Previously, these load balancers supported only certificates with RSA-2048 or ECDSA P-256 key types. With this update, you can now use self-managed certificates with RSA-3072, RSA-4096, and ECDSA P-384 keys.

Key details:

• Supported key types (for self-managed certificates): RSA-2048, RSA-3072, RSA-4096, ECDSA P-256, and ECDSA P-384

• Load balancing coverage for self managed certificates:

  • Certificate Manager SSL certificates: Global and regional load balancing

  • Compute Engine SSL Certificates: Regional load balancing

• Pricing: An additional charge of $0.45 per 1 million connections applies with certificates that use RSA-3072 and RSA-4096 key types. There are no per-connection charges for certificates that use RSA-2048, ECDSA P-256, or ECDSA P-384 key types.

For more information, see the documentation for Supported key types.

This capability is now in General Availability.

General availability: Metrics for all Google Cloud Managed Service for Apache Kafka resources are now available, with some exceptions. Exceptions include the request_count and topic_error_count metrics for Kafka clusters and Kafka Connect connectors. For a list of supported metrics, see Metrics for Cloud Managed Service for Apache Kafka.

Enhanced tagging capabilities for Secret Manager: You can now add tags directly at the time of secret creation. This new feature lets you provide essential metadata for your resources and helps with better organization, cost tracking, and automated policy application from the time a secret is created. In addition to this, tagging for regional secrets is now fully supported, both during secret creation and for existing regional secrets. For more information, see the documentation on tags for global secrets and regional secrets.

Soft-enforced rate limits for modifying secrets and secret versions: We have introduced soft-enforced rate limits for the following operations in Secret Manager:

• AddSecretVersion
• UpdateSecret
• EnableSecretVersion
• DisableSecretVersion
• DestroySecretVersion

Soft enforcement lets us continue serving requests beyond the defined quota as long as our backend systems can comfortably handle the increased load. For details, see the Quotas and limits documentation.

Dynamic Private Service Connect interfaces are available in Preview. You can update VM instances to add or remove dynamic Private Service Connect interfaces without restarting or recreating the instance.

For more information, see Private Service Connect interface types.

VPC Network Peering supports peering connections in consensus mode. This feature is available in Preview. For more information, see Update strategy.

Dataproc Metastore services are now generally available (GA).

Zonal affinity, configured on the backend service of an internal passthrough Network Load Balancer, lets you limit cross-zone traffic, reduce latency, and improve performance, all while maintaining the benefits of a multi-zonal architecture.

Internal passthrough Network Load Balancers support three zonal affinity options that offer varying degrees of preference for routing new connections to eligible backends that are in the same zone as a supported client.

For more information, see Zonal affinity for internal passthrough Network Load Balancers.

This feature is in Preview.

Cloud SQL for SQL Server now offers Active Directory support for write endpoints. For more information, see Write endpoints across forests.

You can now create caches with Anywhere Cache in the us-west3-a, us-west3-b, and us-west3-c zones. For more information about supported locations for Anywhere Cache, see Supported locations.

Vertex AI Agent Engine

Vertex AI Agent Engine Memory Bank is now available in Preview. Memory Bank lets you dynamically generate long-term memories based on users' conversations with your agent.

The following features of policy-based routes are available in General Availability:

• Applying policy-based routes to IPv6 traffic
• Using a next hop that is in a peered VPC network

For more information, see Create policy-based routes.

Preview: You can use future reservations in calendar mode to obtain resources for up to 90 days. By creating a request in calendar mode, you can reserve up to 80 GPU VMs for a future date and time. Then, you can use that capacity to run the following workloads:

• Model pre-training

• Model fine-tuning

• Simulations

• Inference

For more information, see Choose a consumption option.

You can now use your Google Account user credentials to authorize the execution of a data preparation in development. For more information, see Manually run a data preparation in development. This feature is in preview.

Tags data for regional Secret Manager secret usage is available in both the Standard usage cost export and the Detailed usage cost export.

Tags for Global secrets have been available since August 8, 2024. With this update, you can now tag Regional secrets as well.

To learn more about Tags, see Tags overview. To learn about using Tags in your cost data exported to BigQuery, see about tags and query examples with tags.

Cloud Workstations is available in the europe-central2 region (Warsaw). For more information, see Locations.

The Cluster Scheduled Stop feature is available in preview. You can use this feature to stop clusters after a specified idle period, at a specified future time, or after a specified period from the cluster creation or update request.

When upgrading to Windows Server 2022 or newer images on your GCVE private clouds, you can use a combination of Windows Server BYOL for Windows Server 2019 and earlier images and on-demand licensing for Google-provided Windows Server 2022. For more information, see Use a combination of BYOL and on-demand Windows Server licenses from Google.

Dashboards for enhanced visualizations and threat hunting

You can now use the Google SecOps Dashboards to enhance data visualization, investigations, and threat hunting.

Key capabilities include:

• SOAR data availability
• Downloadable reports
• Custom drilldowns
• Markdown widgets
• 51 curated dashboards covering a broad range of security categories and use cases.

For more information, see Dashboards.

Dashboards for enhanced visualizations and threat hunting

You can now use the Google SecOps Dashboards to enhance data visualization, investigations, and threat hunting.

Key capabilities include:

• SOAR data availability
• Downloadable reports
• Custom drilldowns
• Markdown widgets
• 51 curated dashboards covering a broad range of security categories and use cases.

For more information, see Dashboards.

You can now use the Google Cloud console to work with cross-region replication. This feature is Generally Available.

The feature to create Memorystore for Valkey instances in Cluster Enabled and Cluster Disabled modes is now Generally Available.

Share Case Queue Filters

You can now share case queue filters with other users. These filters can be saved with specific criteria, such as assignee roles, and shared with individual users, SOC roles, or all users in your organization for quick access.

For more information, see Apply and save filters.

Share Case Queue Filters

You can now share case queue filters with other users. These filters can be saved with specific criteria, such as assignee roles, and shared with individual users, SOC roles, or all users in your organization for quick access.

For more information, see Apply and save filters.

Document AI now supports Identity and Access Management (IAM) deny policies. These policies allow you to define deny rules that prevent certain principals from using certain permissions to access Google Cloud resources, regardless of the roles they're granted.

For more information, read Deny policy overview and Document AI security and compliance.

Document AI VPC service controls (VPC-SC) integration now supports identity groups.

For more information on setting up VPC-SC identity groups, read Configure identity groups and third-party identities in ingress and egress rules.

Sensitive Data Protection can detect and redact the following object infoTypes in images:

• OBJECT_TYPE/BARCODE
• OBJECT_TYPE/LICENSE_PLATE
• OBJECT_TYPE/PERSON
• OBJECT_TYPE/WHITEBOARD

For more information, see the following:

• Inspect an image for sensitive objects
• Object redaction

The Document AI CDE processor now supports merging the child entities of nested entities that extend across several pages. This is supported in custom extractor model pretrained-foundation-model-v1.5-2025-05-05.

This change is automatically present in existing and newly created processors.

For customers with existing v1.5 processors to make use of this feature, you must relabel the nested entities in different pages.

To learn more about the labeling process, refer to our Label documents documentation.

Vertex AI Agent Garden

Vertex AI Agent Garden now supports filtering by tags.

Vertex AI Search: Search for an exact match (GA)

To search for an exact match, you can enclose your search query in double quotes ("). For example, when you search for "Mary had a little lamb", Vertex AI Search looks for the phrase exactly as it is. It doesn't return search results that contain Mary had lamb, which has missing words; or a little lamb had Mary, which has the words in a different order.

This feature is Generally available when you use the engines.servingConfigs.search method to get search results for custom data, media data, and healthcare data.

The write endpoint feature for Cloud SQL Enterprise Plus edition instances is now generally available (GA). This endpoint is a global domain name service (DNS) name and resolves to the IP address of the current primary Cloud SQL instance that's enabled with private services access.

By using a write endpoint, you can avoid having to make application connection changes after performing a switchover or replica failover operation to test or mitigate a region failure.

For more information, see Connect to an instance using a write endpoint.

The write endpoint feature for Cloud SQL Enterprise Plus edition instances is now generally available (GA). This endpoint is a global domain name service (DNS) name and resolves to the IP address of the current primary Cloud SQL instance that's enabled with private services access.

By using a write endpoint, you can avoid having to make application connection changes after performing a switchover or replica failover operation to test or mitigate a region failure.

For more information, see Connect to an instance using a write endpoint.

Bucket IP filtering for Cloud Storage is now generally available (GA). Bucket IP filtering provides enhanced control over access to your data, allowing you to restrict incoming requests to your Cloud Storage buckets based on their source IP addresses or their Google Cloud Virtual Private Cloud.

Preview: You can reserve GPU VMs that use A4 and A3 Ultra machine types by using future reservations in calendar mode. This feature lets you reserve up to 80 GPU VMs for up to 90 days to obtain capacity for the following workloads:

• Model pre-training jobs

• Model fine-tuning jobs

• High performance computing (HPC) simulation workloads

• Short-term expected increases in inference workloads

For more information, see About future reservation requests in calendar mode.

Generally available: You can create Z3 VMs using smaller machine types, ranging in size from 14 to 88 vCPUs. Also, Z3 now offers -standardlssd and -highlssd predefined machine types. These new machine types have different amounts of Local SSD capacity per vCPU.

When you use Local SSD disks with Z3 VMs, you can receive committed use discounts (CUDs) without needing to attach reservations to your commitments.

For more information, see Storage-optimized machines.

Preview: The general purpose C4 machine series now supports the following machine types on Intel's Xeon 6 processor (Granite Rapids):

• C4 VMs with Titanium Local SSD attached using two new machine types:
  • c4-standard-*-lssd
  • c4-highmem-*-lssd
• Three new bare metal machine types:
  • c4-standard-288-metal
  • c4-highcpu-288-metal
  • c4-highmem-288-metal
• C4 standard, highmem, and highcpu VMs with 144 and 288 vCPUs

To learn more, see the C4 machine series.

For more information, about the attached Local SSD disks, see Machine types that automatically attach Local SSD disks.

Configure storage of Screen Share recordings

You can now configure how long to store Screen Share recordings in your external storage settings. You can also now store Screen Share recordings that originated from the Screen Share adapter.

Administrators: The CRM Comments Creation Details pane at Settings > Operation Management has new Post Cobrowse recording link to CRM record settings. You need to contact Google support to enable Screen Share recordings.

For more information, see Set up external storage for CRMs.

Restrict auto-assignment for email queues

You can now configure email queues so that incoming emails are auto-assigned only during queue operating hours or to agents who are signed in.

Administrators: The Auto assignment dialog at Settings > Queue > Email Edit/View > [queue name] > Automatic assignment > Configure has two new checkboxes.

For more information, see Email auto assignment.

Salesforce: New closed record options for scheduled calls

If you've integrated Google Cloud CCaaS with the Salesforce CRM, you can configure how your instance handles scheduled calls for closed records. Here are the configuration options that are available with Salesforce integrations:

• Don't look up record status. Your instance doesn't look up record status and makes the scheduled call regardless of whether the record is open or closed.

• Look up record status and reopen closed records. Your instance looks up record status before making a scheduled call and reopens records that are closed.

• Look up record status and cancel scheduled calls for closed records. Your instance looks up record status before making a scheduled call and cancels scheduled calls for records that are closed.

Administrators: The CRM Record Creation Details pane at Settings > Operation Management has new settings at Closed record options when initiating an API-scheduled call.

For more information, see Schedule calls with Salesforce.

Okta: Version 9.0

• The following new action has been added:

  • Send SSF to Okta

Support for AppGroups in Abuse Detection attributes

Abuse Detection incidents and detected traffic now show information on AppGroups and AppGroup apps when the AppGroup is part of the request or traffic.

Note: This functionality is not available in Apigee hybrid at this time.

For usage information, see the Abuse Detection documentation.

You can now update a Cloud KMS encryption key by updating the table with the same key. This feature is generally available (GA).

You can use the @@location system variable to set the location in which to run a query. This feature is generally available (GA).

BigQuery now supports the following Apache Hadoop migration features in Preview:

• Use the dwh-migration-dumper tool to migrate the metadata necessary for a Hadoop permissions and data migration.
• Migrate permissions from Apache Hadoop, Apache Hive, and Ranger HDFS to BigQuery.
• Migrate tables from a HDFS data lake to Google Cloud.

(New guide) GraphRAG infrastructure for generative AI using Vertex AI and Spanner Graph: Shows how to design infrastructure for GraphRAG-capable generative AI applications in Google Cloud by using Vertex AI and Spanner Graph.

VMware Engine single-node private clouds, created after June 4th 2025, will no longer delete after 60 days. All single-node private clouds will have no time limit. Single-node private clouds have limitations to SLA, patching, and upgrades.

Policy Simulator for Organization Policy is now generally available (GA).

Policy Simulator for Organization Policy is now generally available (GA).

Plugins for Cloud Load Balancing help you insert WebAssembly (Wasm) code in a fully managed serverless environment directly into the data path of most Cloud Load Balancing Application Load Balancers. This feature is in General Availability.

The global endpoint is generally available (GA) for Anthropic's Claude Opus 4. For details, see Global endpoint.

For new deployments, the URL Fetch API validates the certificate of the host it contacts by default.

For new deployments, the URL Fetch API validates the certificate of the host it contacts by default.

You can now create and manage scheduled notebooks using the Schedule details pane in BigQuery Studio. This feature is generally available (GA).

Using a fully qualified domain name (FQDN) forwarding target is available for outbound DNS forwarding in GA.

Cloud HSM for Google Workspace now lets you use Cloud HSM keys for client-side encryption (CSE) to protect sensitive workloads in Google Workspace. For more information about Cloud HSM for Google Workspace, including how to get started, see Onboard to Cloud HSM for Google Workspace.

You can apply maximum instance configuration at the service level (in Preview).

Preview: You can consume reservations with Colab Enterprise runtimes. Reservations of Compute Engine zonal resources help you gain a high level of assurance that your runtimes have the necessary resources to run. For more information, see Use reservations with Colab Enterprise.

Generally available: You can now modify licenses attached to your disks. Previously, licenses on disk resources were immutable. You had to delete and recreate disks, or engage our support team to change licenses.

This feature provides greater flexibility for managing your disk licenses. You can now:

• Append, remove, replace, and view the history of license updates.
• Perform in-place license upgrades, such as Ubuntu to Ubuntu Pro, using the gcloud CLI and REST.
• Switch from PAYG to BYOS billing models.
• Review license changes and restrictions and append a RHEL ELS license to a newer version.

For more information on how to manage licenses, see Manage licenses.

Datastream is now available in the northamerica-south1 (Mexico) region. For the list of all available regions, see IP allowlists and regions.

Custom Extractor model pretrained-foundation-model-v1.5-2025-05-05 is in General Availability (GA) and has fine-tuning available for the US and EU.

From version v1.4 and later, we will use a new quota for online processing called Number of online process document pages per minute per processor type and model version. This quota will be enforced at a per-page and per-foundation model level. There will be no change to the batch processing quota.

These can be enabled in the console when creating labels and by using the DocumentSchema.EntityType.

For more information, read Managing processor versions.

The Fast Dev Mode Transition feature is out of Labs and is now generally available. The Fast Dev Mode Transition feature improves the performance of Development Mode on your instance by loading LookML projects in read-only mode until a developer clicks the Create Developer Copy button for the project. Note: This item was added on July 8, 2025.

The Fast Dev Mode Transition feature is now available for Looker (Google Cloud core). The Fast Dev Mode Transition feature improves the performance of Development Mode on your instance by loading LookML projects in read-only mode until a developer clicks the Create Developer Copy button for the project. Note: This item was added on July 8, 2025.

You can download risk reports as PDFs. Risk reports help you understand the results of the attack path simulations (virtual red teaming) that Security Command Center runs. This feature is in Preview and is available for customers on the Enterprise or Premium service tiers. For more information, see Risk reports overview.

The following Virtual Machine Threat Detection detectors are in General Availability.

• Defense Evasion: Unexpected ftrace handler
• Defense Evasion: Unexpected interrupt handler
• Defense Evasion: Unexpected kernel modules
• Defense Evasion: Unexpected kernel read-only data modification
• Defense Evasion: Unexpected kprobe handler
• Defense Evasion: Unexpected processes in runqueue
• Defense Evasion: Unexpected system call handler

Spanner supports the following new client-side metrics to the Spanner API frontend (AFE) and Google frontend (GFE) for Java and Go applications:

• AFE connectivity error count
• AFE latencies
• GFE connectivity error count
• GFE latencies

These metrics can be used with server-side metrics to enable faster troubleshooting of performance and latency issues. For more information, see Client-side metrics descriptions.

To troubleshoot or understand your Spanner queries better, you can download and save your query execution plan as a JSON file. You can now use the content of this file to see a visualization of the query execution plan in Spanner Studio. For more information, see Take a tour of the query plan visualizer.

Preview stage support for the following integration:

• FleetPackage API

Vertex AI online inference now offers Preview support of PSC service automation that can automatically create PSC endpoints for dedicated private endpoints. For more information, see Create the online inference endpoint with PSC automation.

Vertex AI now offers GA support of Private Service Connect Interface and includes Private DNS Peering. For more information, see Use Private Service Connect interface for Vertex AI Training.

Private Service Connect interface (PSC-I) support for ML pipeline runs in Vertex AI Pipelines is now generally available. PSC-I is recommended for private connectivity because it reduces the chance of IP exhaustion, allows for transitive peering, and includes Private DNS Peering.

For more information, see Configure Private Service Connect interface for a pipeline.

You can backdate the not_before_time of certificates by specifying the backdate_duration field within the issuance policy of your CA Pool. This new optional field in the issuance policy allows you to control the not_before_time of all certificates issued from a given CA Pool.

If backdate_duration is not set: Certificates are issued with a not_before_time equal to the current issuance time.

If backdate_duration is set: Certificates are issued with a not_before_time equal to the issuance time minus the specified backdate_duration. The not_after_time automatically adjusts to maintain the requested certificate lifetime.

The following resource types are now publicly available through the ExportAssets, ListAssets, BatchGetAssetsHistory, QueryAssets, Feed, and Search (SearchAllResources, SearchAllIamPolicies) APIs.

• Cloud Spanner
  • spanner.googleapis.com/InstancePartition

Generally available: You can specify a custom ephemeral external IPv6 address when creating an instance. For more information, see Create instances that use IPv6 addresses.

Quality AI offers the agent engagement platform where contact center personnel can visualize agent performance data, including an AI-generated summary. The agent engagement platform also identifies areas where an agent excels and needs help compared to their peers.

Qualilty AI offers agent assessments based on individual conversations. Human evaluators can also add notes to these assessments with feedback for the agent.

Quality AI offers sampling rules that filter conversations to reduce the workload for human evaluators, saving contact centers time and money.

Conversational Insights offers a devkit to help developers and maintainers perform a wide range of actions, including the following:

• Import one or more conversations with metadata.
• Transcribe mono audio files.
• Create recognizers with STT V2.
• Transform transcript data formats from Genesys Cloud or AWS.
• Change Conversational Insights global settings.

Gemma 3n models are now available through Model Garden.

Multimodal datasets are now available in preview. For more information, see Multimodal datasets.

Cloud Armor's Hierarchical security policies facilitate centralized control, enhanced consistency, operational efficiency, and effective delegation of security policy management in Preview.

Gemma 3n models are now available through Model Garden.

Private Service Connect service connectivity automation periodically retries endpoint create or delete operations that fail due to errors. This feature is available in General Availability. For more information, see Automatic retries for endpoint failures.

Vertex AI Search: gemini-2.5-flash/answer_gen/v1 model

You can generate answers with the Gemini 2.5 Flash (gemini-2.5-flash) model. This model is tuned to address context-based question and answering tasks.

For more information, see Answer generation model versions and lifecycle.

You can now use the use the PARTITION BY clause of the CREATE VECTOR INDEX statement to partition TreeAH vector indexes. Partitioning enables partition pruning and can decrease I/O costs. This feature is in preview.

BigQuery search indexes provide free index management until your organization reaches the limit in a given region. You can now use the INFORMATION_SCHEMA.SEARCH_INDEXES_BY_ORGANIZATION view to understand your current consumption towards that limit, broken down by projects and tables. This feature is generally available (GA).

In typical HTTPS communication, neither the load balancer nor the backend verify each other's identity, assuming that they are within a secure perimeter and can be trusted. However, when perimeter security needs reinforcement or communication extends beyond the perimeter, backend mTLS becomes essential. Backend mTLS ensures secure communication by requiring both the load balancer and the backend to mutually verify their identities.

With backend authenticated TLS, the load balancer verifies the backend server's certificate by checking its chain of trust, thereby confirming the backend's identity. Conversely, with backend mTLS, the backend server verifies the client certificate presented by the load balancer. Together, these mechanisms enable backend mTLS, ensuring that both parties validate each other's identity.

Backend mTLS complements frontend mTLS, which is already generally available (GA).

For details, see the following:

• Backend mTLS overview
• Set up backend authenticated TLS
• Set up backend mTLS

This capability is in General Availability for global external Application Load Balancers.

Dataflow now supports an automated parallel update workflow for streaming jobs. This feature helps minimize disruption by launching a new replacement job that runs in parallel with the existing job. After a duration of time you specify, the old job is automatically drained.

For more information, see Run parallel pipelines.

Conversational Agents data stores: Conversational Agents now supports AlloyDB, Bigtable, Firestore, Spanner and Cloud SQL data store sources as public GA features. Microsoft Entra ID source has now private GA support.

Conversational Agents data stores: You can now create some data store types directly within the Conversational Agents console rather than needing to use AI Applications. See the data store creation documentation for details.

Looker connector enhancements

The Looker connector can now connect to a private IP (private services access) only Looker (Google Cloud core) instance or to a private IP (Private Service Connect) Looker (Google Cloud core) instance using the Looker instance ID.

Static routes for Network Connectivity Center are available in public preview.

You can use static routes to define the next hop along the path that network traffic takes to reach a given destination. For more information about using static routes with Network Connectivity Center, see the Static routes overview.

General availability support for the following integration:

• Parameter Manager

M130 release

The M130 release of Vertex AI Workbench instances includes the following:

• Updated the Dataproc JupyterLab plugin to version 0.1.87.
• Added the BigQuery JupyterLab plugin, version 0.0.1.
• The GOOGLE_CLOUD_REGION environment variable is now set by default.

App Hub supports resources from the following sources in Preview:

• Dataproc Metastore Service
• Vertex AI Dataset
• Vertex AI Featurestore
• Vertex AI MetadataStore
• Vertex AI Model

Cloud Run worker pools are now available (Preview). Worker pools are specifically designed for non-request workloads.

Use agent mode in Gemini Code Assist Standard and Enterprise insiders channel for VS Code (Preview)

Stay in control with interactive reviews and approvals.

You can use Gemini Code Assist chat in agent mode to complete complex, multi-step tasks and goals with complete control over every change. Before making any modifications, the agent will present a plan for your review. You can edit, ask for changes, approve, or deny any suggested changes. This collaborative approach combines the power of AI with your expertise, resulting in better code and a more efficient workflow.

To get started with agent mode, see Use agentic chat as a pair programmer.

Use multi file editing in Gemini Code Assist Standard and Enterprise insiders channel in agent mode for VS Code (Preview)

Say goodbye to single-file edits and hello to project-wide changes.

With multi-file edits in agent mode, the agent can make concurrent changes across your entire codebase in response to a single prompt. This powerful new capability streamlines large-scale refactoring, feature implementation, and bug fixes. Simply describe the changes you need, and the agent will intelligently identify and modify all relevant files, saving you time and effort. You will also have the option to undo changes to local files in case you want to revert the changes to an earlier state.

Use full project context in Gemini Code Assist Standard and Enterprise insiders channel in agent mode for VS Code (Preview)

Smarter, more accurate code suggestions with full project awareness.

Agent mode has a comprehensive understanding of your entire project. The agent analyzes your whole codebase and requests files and folders as needed based on your goals. Full project context lets the agent create more accurate and context-aware code completions, suggestions, and refactorings. This deeper understanding of your project's architecture, dependencies, and coding patterns means you get higher-quality, more consistent code with less effort.

The C4D machine series is generally available in GKE. The following version requirements apply:

• Standard clusters:
  • Manual node creation: GKE version 1.30 and later.
  • Node auto-provisioning and cluster autoscaler with Confidential GKE Nodes and compact placement: GKE version 1.32.3-gke.1717000 and later.
• Autopilot clusters, including compact placement:
  • C4D machine types without Titanium SSD: GKE version 1.33.0-gke.1439000 and later.
  • C4D machine types with Titanium SSD: GKE version 1.33.1-gke.1171000 and later.

You can use the C4D machine series with Confidential GKE Nodes and in compact placement policies in Autopilot and Standard clusters.

For more information, see C4D machine series.

The C4D machine series is generally available in GKE. The following version requirements apply:

• Standard clusters:
  • Manual node creation: GKE version 1.30 and later.
  • Node auto-provisioning and cluster autoscaler with Confidential GKE Nodes and compact placement: GKE version 1.32.3-gke.1717000 and later.
• Autopilot clusters, including compact placement:
  • C4D machine types without Titanium SSD: GKE version 1.33.0-gke.1439000 and later.
  • C4D machine types with Titanium SSD: GKE version 1.33.1-gke.1171000 and later.

You can use the C4D machine series with Confidential GKE Nodes and in compact placement policies in Autopilot and Standard clusters.

For more information, see C4D machine series.

Flexible shielding helps you overcome some limitations of default origin shielding by letting you configure a single, specific geographic region for origin shielding, typically selected to be near your centralized origin. This feature is in Preview.

The version upgrade feature for Memorystore for Valkey is now Generally Available (GA).

The CZECHIA_PERSONAL_ID_NUMBER infoType detector is available in all regions. For more information about all built-in infoTypes, see InfoType detector reference.

You can use Data Boost to analyze your Bigtable data with BigQuery without impacting the performance of the clusters that handle your application traffic. This feature is generally available (GA).

New, enhanced forecasting model for increased accuracy in cost reports

Cloud Billing forecasts now better account for seasonality trends, data irregularities, and missing data, using an enhanced forecasting model that leverages AI to factor in various scenarios, such as the following:

• Intelligent handling of transient effects caused by known business events - for example, a new workload migration causing a usage spike.
• Deeper understanding of seasonality - for example, various recurring patterns, such as daily, weekly and monthly cycles in your cloud spend; or for retailers, increases in usage during holiday seasons.
• Adapting to trends to remain relevant in changing environments - for example, new AI spend.

These enhancements, powered by our new machine learning engine, translate to increased forecasting accuracy. By capturing complex trends, multiple seasonalities, and handling data anomalies more intelligently, you'll see a marked improvement in the precision of your cost forecasts.

For more information about the forecasted costs in reports, see View you forecasted costs.

A new region is now available for Cloud Run GPUs: us-east4.

MySQL 8.4.4 is upgraded to MySQL 8.4.5. For more information, see the MySQL 8.4.5 Release Notes.

Cloud Armor supports organization-scoped address groups for security policies in Preview.

The following feature is generally available for Looker reports:

• The Looker connector can now connect to a private IP (private services access) only Looker (Google Cloud core) instance or to a private IP (Private Service Connect) Looker (Google Cloud core) instance using the Looker instance ID.

You can directly connect and interact with your Spanner database using the Spanner CLI, an interactive shell for Spanner that is built into the Google Cloud CLI. You can use the Spanner CLI to start an interactive session and automate SQL executions from the shell or an input file. This feature is available in Preview. For more information, see Spanner CLI quickstart.

Addition of AppGroup-specific Analytics dimensions for Custom Reports

This release introduces two new AppGroups Analytics dimensions: AppGroup Name and AppGroup App Name.

Use these dimensions with custom reports and report jobs to group metrics by a specific AppGroup or a specific app within an AppGroup.

For additional information see Analytics dimensions and Creating and managing custom reports.

This release adds the Export feature to the Apigee UI in the Cloud console. You can now export publishing data for developers, apps, or API products as a comma-separated values (CSV) file or JSON file.

Documentation: Exporting publishing data

You can now use the Apache Iceberg REST catalog in BigLake metastore to create interoperability between your query engines by allowing your open source engines to access Iceberg data in Cloud Storage. This feature is in Preview.

Colab Enterprise notebooks in BigQuery let you do the following in Preview:

• Explain code with Gemini assistance
• Fix and explain errors with Gemini assistance

Cloud CDN lets you add custom code to the request processing path of global external Application Load Balancers by using Service Extensions edge extensions. These extensions help you implement customizations in the request path pre-cache, which can influence how content is cached and the backend service is selected on the Application Load Balancer. This feature is in Preview.

For more information, see Use Service Extensions for edge computing.

Improved the startup times of Airflow workers for environments that have a large number of custom PyPI packages installed.

This feature was announced previously and is gradually rolling out over several releases. In this release, it's available in asia-east2, asia-northeast3, europe-central2, europe-west9, me-central1, me-west1, northamerica-northeast1, northamerica-northeast2, northamerica-south1, and us-west1 re

Interconnect connection groups and VLAN attachment groups are Generally available. You can use resource groups to communicate your intended level of reliability, and to receive feedback on how your Cloud Interconnect resources meet that intended level of reliability.

For more information, see Resiliency and SLA options.

You can create a secure tag at the organization level and bind its value to all virtual machine (VM) instances across that organization, instead of applying tags to instances within a specific network. For more information, see Secure tags for firewalls. This feature is available in Preview.

The Cloud Storage Cloud Audit Logs have expanded support to include error scenario coverage and produce a more comprehensive error message with code, error messages, and details, in an easy to understand format. The Gemini Cloud Assist (GCA) service can then easily analyze the log and provide tailored recommendations on how to mitigate issues as they arise. Before this enhancement, error logs were generated for only a specific set of scenarios, and the status field solely contained the gRPC error code without any additional information. To learn more about Cloud Audit Logs, see Cloud Audit Logs overview.

Added NVIDIA 570.133.20 vGPU driver.

Added a kernel patch to address bcache latency.

Added a kernel patch to address bcache latency.

Added a kernel patch to address bcache latency.

Datastream now supports BigLake Iceberg tables as a destination both in the Google Cloud console and the Datastream API.

For more information, see the Datastream documentation.

Edge extensions help you manipulate request headers early in the request processing lifecycle of global external Application Load Balancers to influence caching and routing decisions. This feature is in Preview.

Learn how to configure an edge extension.

The IRS 1075 control package now supports the following products:

• Access Transparency
• Backup for GKE
• Cloud Healthcare API
• Cloud OS Login API
• Cloud Workstations
• Conversational Agents
• Conversational Insights
• Dataplex Universal Catalog
• Document AI
• Essential Contacts
• Eventarc
• External passthrough Network Load Balancer
• Generative AI on Vertex AI
• Google Agentspace
• Google Cloud Armor
• Google Cloud NetApp Volumes
• Google Security Operations SOAR
• Identity and Access Management (IAM)
• Infrastructure Manager
• Integration Connectors
• Internal passthrough Network Load balancer
• Jurisdictional Cloud Console
• Looker (Google Cloud core)
• Organization Policy Service
• Regional external Application Load Balancer
• Regional external proxy Network Load Balancer
• Regional internal Application Load Balancer
• Regional internal proxy Network Load Balancer
• Storage Transfer Service
• VPC Service Controls
• Virtual Private Cloud (VPC)

The following resource types are now publicly available through the ExportAssets, ListAssets, BatchGetAssetsHistory, QueryAssets, Feed, and Search (SearchAllResources, SearchAllIamPolicies) APIs.

• Security Command Center
  • websecurityscanner.googleapis.com/ScanConfig

Conversational Insights offers sentiment analysis in GA. Sentiment analysis determines the mood within a conversation and assigns a score: positive, neutral, or negative.

Starting with version Migrate Connector 2.7, throttling is supported for a second NIC.

A new free trial creation work flow makes it easier to start your Spanner free trial. With a free trial instance, you can learn and explore Spanner for 90 days at no cost. You can create relational (GoogleSQL and PostgreSQL) databases and deploy NoSQL models (Spanner Graph, Vector search, and Full-text search) in a single, fully managed database. For more information, see Spanner free trial instances overview.

We've increased the maximum file size for online processing requests from 20 MB to 40 MB. This applies to all types of processors.

For more information, see the Document AI limits page.

Content Hub

This feature is currently in Preview.

The new Content Hub page offers a centralized experience for managing all your Google SecOps content needs. On this page, you can do the following:

• Onboard Google SecOps content using content packs for top data sources
• View and manage native dashboards.
• Access and configure search queries.
• View, filter, and review curated detections rule logic.
• Configure response integrations.
• Install and run power ups.

For more information, see Google SecOps Content Hub.

Product Centric Feed Management

This feature is currently in Preview.

You can now configure multiple log-type feeds for the same product type on a single page. This new product-led experience simplifies the feed configuration flow and provides additional in-product guidance. For more information, see Configure feeds by product.