Gemini for Google Cloud release notes

Gemini Cloud Assist is now supported within VPC Service Controls perimeters. For more information, see VPC Service Controls supported services.

Gemini 3.5 Flash is generally available

Gemini 3.5 Flash is now generally available to Gemini Code Assist users in VS Code and IntelliJ. You can use this model for agent mode, chat, and code generation.

Agent mode bug fix in Cloud Workstations

Fixed a bug that prevented agent mode from working on Cloud Workstations.

Bug fixes in IntelliJ

Various bug fixes and minor product enhancements.

Model Context Protocol support

MCP support is available in Private Preview. To request access, contact your Google Cloud account team. You can interact with Gemini Cloud Assist agents from various surfaces, including third-party client agents and IDEs, using the Model Context Protocol (MCP). For more information, see Integrating with MCP.

Proactive agents for issue investigation

Proactive agents are available in Private Preview to Premium Support customers. Enable Gemini Cloud Assist to autonomously investigate issues triggered by Cloud Alerting policies or cost anomalies in the background. These investigations don't modify or make any changes to your environment. Results and insights, such as root cause analysis for alerts or cost spike drivers, are delivered via Eventarc. You can view these results in the Google Cloud console. This feature requires administrator enablement and configuration of an agent identity. For more information, see Configure proactive agents.

Page context awareness

Page context awareness is available in Public Preview. Gemini Cloud Assist automatically uses the context of the content currently visible on your Google Cloud console page to provide more relevant and accurate responses to your prompts. For more information, see Manage page context sharing.

Enhanced agent administration controls

Enhanced agent administration controls are available in Public Preview. Administrators have the following options in the Cloud Assist settings panel:

• Web grounding settings: Choose how Gemini Cloud Assist uses Google Search to ground its responses, with options for either the most comprehensive results or stricter data residency compliance. For more information, see Configure web grounding.
• Custom instructions: Define persistent system instructions to customize the agent's persona, response formatting, and awareness of organizational standards or business context for all users within the project. For more information, see Set custom instructions.

Bug fixes in IntelliJ

Various bug fixes and minor product enhancements.

geminicloudassist API automatically enabled for Gemini Cloud Assist chat users

As of April 16th, 2026, the geminicloudassist.googleapis.com API has been automatically enabled on projects that meet all of the following criteria:

• Had used Gemini Cloud Assist chat in the prior 60 days.
• Had the cloudaicompanion.googleapis.com API enabled on April 16, 2026.
• Did not have the geminicloudassist.googleapis.com API enabled on April 16, 2026.

The Gemini Cloud Assist chat functionality that was previously served by cloudaicompanion.googleapis.com is now served by geminicloudassist.googleapis.com, and both APIs are dependencies to use Gemini Cloud Assist. This automatic API enablement ensures that users have access to the same functionality without any loss of service.

Quota issues fixed in VS Code

Various bug fixes related to quota issues.

Update for Gemini Cloud Assist investigations

From April 10, 2026, creating, running, and editing Gemini Cloud Assist investigations will only be available to users with a Premium Support contract or who have requested access through their account team. If you ran an investigation prior to April 10, 2026, the results of the investigation continue to be available to you in the Google Cloud console.

Bug fixes in IntelliJ

Various bug fixes and minor product enhancements.

VPC Service Controls support for Gemini Cloud Assist investigations deprecation

Starting April 13, 2026, Gemini Cloud Assist investigations will no longer be supported within VPC Service Controls perimeters. You do not need to remove the Gemini Cloud Assist API (geminicloudassist.googleapis.com) from your Restricted Service configurations, but if you keep it there, access to it will still be blocked within the perimeter. To continue using this feature, you must make requests to the service from outside any VPC Service Controls perimeter. For more information, see the Deprecated VPC Service Controls support for Gemini Cloud Assist investigations page.

Agent mode logs now attributed to Gemini Code Assist

VS Code Gemini Code Assist 2.77.1 now attributes agent mode logs to Gemini Code Assist. In previous versions, agent mode logs are being attributed to Gemini CLI instead of Gemini Code Assist. This discrepancy is resolved in the latest release, and we recommend that you update to version 2.77.1 or higher to ensure your usage metrics are correctly reported.

Custom IAM roles permission update for Gemini Cloud Assist

Gemini Cloud Assist has replaced the cloudaicompanion.instances.completeTask IAM permission with geminicloudassist.agents.invoke. If you have access to Gemini Cloud Assist through a custom IAM role, you must update the role to continue having access. For more information, see the deprecated IAM permissions document.

Custom IAM roles permission update for Gemini Cloud Assist

On April 8, 2026, Gemini Cloud Assist is replacing the cloudaicompanion.instances.completeTask IAM permission with geminicloudassist.agents.invoke. Updates to standard IAM roles will be done automatically, but if you have access to Gemini Cloud Assist through a custom IAM role, you must update the role before April 8, 2026 to ensure continued access. For more information, see the deprecated IAM permissions document.

Bug fixes in IntelliJ

Various bug fixes and minor product enhancements.

Gemini 3.1 Pro and 3.0 Flash are available (Preview)

Gemini 3.1 Pro and Gemini 3.0 Flash are now available to Gemini Code Assist users in VS Code and IntelliJ, in Preview. You can use these models for agent mode, chat, and code generation.

File outline and finish changes features in VS Code general availability

The file outline and finish changes features are now available in VS Code and generally available (GA).

File outline feature in IntelliJ general availability

The file outline feature is now generally available (GA).

Finish changes feature in IntelliJ general availability

The finish changes features is now generally available (GA).

VPC Service Controls for Gemini Cloud Assist investigations is now available in Preview.

Gemini Cloud Assist documentation migration

Gemini Cloud Assist release notes are now published here. If you are looking for Gemini Cloud Assist release notes prior to January 2026, see the Gemini for Google Cloud release notes.

VS Code model selection issue fixed as of version 2.63.1

The model selection issue affecting free tier customers using Gemini Code Assist releases 2.56 and later is fixed as of version 2.63.1.

Known issue for the model selection feature in VS Code Gemini Code Assist

The VS Code Gemini Code Assist model selection feature incorrectly appears for users of the free tier of Gemini Code Assist for individuals.

• The feature shouldn't appear for such users and does not function for them.

Finish Changes in IntelliJ (Preview)

The finish changes feature acts as an AI pair programmer that observes your in-progress work and completes the job, eliminating the need to write complex prompts. The core strength of the finish changes feature lies in its flexibility, allowing you to use a mix of input styles, such as pseudocode, #TODOs, or half-written code. This combination minimizes distraction and ensures you remain in control of the high-level design.

Outline in IntelliJ (Preview)

The outline feature automatically creates AI-assisted documentation by generating short, English summaries of code blocks within the Outline tab of the Gemini Code Assist plugin. Outlines help developers to achieve rapid understanding and to focus on abstraction, without getting bogged down by syntactical details.

You can change the settings to disable automatic outline generation, which instead lets you generate outlines of files manually.

Model selection for VS Code Gemini Code Assist

The following VS Code Gemini Code Assist users can now manually select the model used by Gemini Code Assist:

• Gemini Code Assist Enterprise users
• Gemini Code Assist Standard users
• Gemini Code Assist for individuals, if you have a Google AI Pro or Ultra subscription

VS Code server connection issue fixed as of version 2.58.1

The server connection issue affecting Gemini Code Assist 2.57 and 2.58 releases is fixed as of version 2.58.1

Known issue in VS Code 2.57 and later

The Gemini Code Assist 2.57 and later releases for VS Code have a known issue that prevents the extension from connecting to the server if you are using a gnc c runtime version that's earlier than 2.36.

• You can permanently fix this issue by upgrading your gnu c runtime version to 2.36 or higher.

• Alternatively, you can downgrade the Gemini Code Assist VS Code extension to version 2.56 until a fix is released.

Chat performance metrics

Gain deeper insights into Gemini for Google Cloud's chat performance with metrics for chat acceptance rate and lines of chat accepted, including interactions in Agent mode and non-agentic chat. Combine these metrics with existing code metrics to calculate total acceptance and total lines of code accepted, providing a complete picture of Gemini's performance. For more information, see View Gemini for Google Cloud logs.

Code customization available in agent mode and Gemini CLI (Preview)

Code customization is now supported when using Gemini CLI and agent mode in Gemini Code Assist.

Inline diff is generally available (GA) for IntelliJ

Inline diff is generally available for IntelliJ Gemini Code Assist. With the inline diff view, chat generated code suggestions are visually highlighted directly in your editor. This feature can boost 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 also change the diff view settings to have a side-by-side diff window instead of the inline diff.

Persistent memory for Gemini Code Assist on GitHub (Preview)

• Gemini Code Assist on GitHub now supports persistent memory, which stores your previous interactions with Gemini Code Assist on GitHub so that it has context during your future interactions.

• For users of the enterprise version, persistent memory is enabled from the connection settings within the Google Cloud console. For users of the consumer version, persistent memory is enabled from the Gemini Code Assist site. For more information, see manage configuration files across multiple repositories.

Gemini Code Assist tools are no longer available and are replaced by agent mode (Preview)

Effective October 14, 2025, Gemini Code Assist tools are no longer available. This completes the deprecation process that was announced on October 2, 2025.

Using the @ symbol followed by the name of a tool no longer connects to Gemini Code Assist tools.

To connect to external services, use Model Context Protocol (MCP) servers in Gemini Code Assist agent mode.

For more information, see:

• Configure MCP servers
• Gemini feature deprecations

Enterprise version of Gemini Code Assist on GitHub (Preview)

• Gemini Code Assist on GitHub provides Gemini-powered reviews of pull requests in your GitHub repositories.

• The enterprise version enables consolidated control of Gemini Code Assist reviews across multiple repositories, provides increased quotas, supports GitHub Enterprise Cloud and GitHub Enterprise Server, and operates under the Google Cloud Terms of Service.

Security, privacy, and compliance for Gemini in BigQuery

Security, privacy, and compliance for Gemini in BigQuery details how customer data is protected and processed by Gemini in BigQuery.

Next Edit Predictions in IntelliJ (Preview)

Next Edit Predictions, which predicts the next code suggestions throughout the code file that you're currently in, are now available in IntelliJ Gemini Code Assist, in Preview.

You can cycle through multiple suggestions, dismiss suggestions, or ignore them and continue coding.

Enable Next Edit Predictions in your IDE to get started.

Chat code suggestion preview in IntelliJ

Chat code suggestions are displayed in a preview block by default with IntelliJ Gemini Code Assist, improving the readability of generated chat responses.

Gemini Code Assist tools are deprecated and replaced by agent mode

Gemini Code Assist tools are deprecated and will be removed on October 14, 2025. Gemini Code Assist tools are replaced by Gemini Code Assist agent mode. After October 14, 2025, using the @ symbol followed by the name of a tool won't connect to Gemini Code Assist tools.

Agent mode in Gemini Code Assist can connect to external services using Model Context Protocol (MCP) servers. For more information, read Configure MCP servers.

Add code snippets to the chat context

You can now select, attach, and direct Gemini to focus on code snippets with IntelliJ Gemini Code Assist. Code snippet selection enables discrete analysis of smaller code blocks instead of entire files.

Next Edit Predictions in VS Code (Preview)

Next Edit Predictions, which predicts the next code suggestions throughout the code file that you're currently in, are now available in VS Code Gemini Code Assist, in Preview.

You can cycle through multiple suggestions, dismiss suggestions, or ignore them and continue coding.

Enable Next Edit Predictions in your IDE to get started.

Cloud Hub Optimization and Cost Explorer integrations

Gemini Cloud Assist can now answer questions about your resource cost and utilization. To learn more about what permissions you need to enable this functionality, see the Optimize resource costs and utilization setup and the Optimize costs with Cost Explorer setup.

Copy sign-in link

As an alternative to clicking Sign in when signing into your Google Account to use IntelliJ Gemini Code Assist, you can click Copy link and manually paste the link in your browser's URL bar.

Access saved prompts in the Prompt Library

You can access saved prompts in the Prompt Library when you type @ in Gemini chat and select the prompt you want.

To view the Prompt Library settings, go to Settings > Tools > Gemini > Prompt Library.

Inline diff is generally available (GA) for VS Code

Inline diff is generally available for VS Code Gemini Code Assist. 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 also change the diff view settings to have a side-by-side diff window instead of the inline diff.

Revert to a checkpoint in IntelliJ (GA)

Reverting to a checkpoint for chat code suggestions with IntelliJ Gemini Code Assist is now generally available. Reverting to a checkpoint lets you revert affected source files to a point before any code suggestions were applied.

See code customization status directly in IDE

You'll now get a one-time notification the first time code customization is enabled for you, so you know right away that the feature is active. Additionally, you can now check the status of code customization directly by clicking on the Gemini icon in your IDE. This makes it simple to confirm that the feature is configured correctly and ready to go. For more information, see Code customization overview.

Performance and stability improvements for VS Code

Numerous performance and stability improvements have been made to the Gemini Code Assist extension, further reducing extension crashes and related adverse functional events.

Edit a prior prompt in VS Code

You can edit a prior prompt in Gemini Code Assist and receive a regenerated response based on your edited prompt.

Introducing Release Channels in Gemini Code Assist

We're rolling out Release Channels in Gemini Code Assist Standard and Enterprise to give you more control over the features and models you access, helping us deliver new capabilities faster. This means direct access to our latest features, minimizing the need for sign-ups.

You can choose between the Preview Channel for early access to cutting-edge features (no SLA) or the GA Channel for stable, fully supported features with a standard Google Cloud SLA. Project leads and administrators can easily configure these settings at the Google Cloud Platform project level within the Admin for Gemini settings, making it simple to opt in your entire team to the Preview channel if desired. Note that at launch, the two release channels have identical features. For more information, see Configure Gemini Code Assist release channels.

Delete prompt and response pair in VS Code

You can delete your prompt and Gemini's response to that prompt in your chat with Gemini Code Assist. This works as an alternative to deleting your entire chat history, allowing you to remove a single prompt and response within a chat, while maintaining the rest of your chat history with Gemini Code Assist.

Regenerate a prompt response in VS Code

You can regenerate a prompt response to your most recent prompt, where Gemini Code Assist re-evaluates your prompt and provides a new response.

Deploy apps to Cloud Run in Gemini Code Assist agent mode (Preview)

Leverage the power of Gemini CLI extensions in VS Code Gemini Code Assist agent mode with the /deploy custom slash command. The /deploy command lets you deploy your existing web application to Cloud Run directly from agent mode. /deploy takes care of what used to be a multi-step process of building, containerizing, pushing, and configuring, and then returns a public URL for your live application.

To get started using the /deploy command, create a web application in your workspace, install the Cloud Run MCP server, turn on agent mode, and then type /deploy. Your app is deployed in minutes and accessible through a public URL. For more information, see Use commands.

Regenerate a prompt response in IntelliJ

You can regenerate a prompt response to your most recent prompt, where Gemini Code Assist re-evaluates your prompt and provides a new response.

Add terminal output to context in IntelliJ

You can attach terminal output to the chat context with IntelliJ Gemini Code Assist. You can now ask Gemini Code Assist questions about terminal commands and output.

Monitor Gemini Code Assist usage

You can now monitor your organization's usage of Gemini Code Assist with a dashboard that is automatically available when you enable and use Gemini Code Assist. The dashboard includes some of the most important metrics, giving you a quick way to view aggregated usage data. For more information, see Monitor Gemini Code Assist usage.

Mention a remote repository to prioritize context

You can now start your prompt with the @ symbol and select a specific remote repository from a list to ensure Gemini prioritizes that context. This feature is designed to give you more relevant and precise suggestions by explicitly directing Gemini's focus to the codebase you care about most, allowing you to get more tailored and accurate results. For more information, see Code customization overview.

Create and manage multiple chats in IntelliJ

You can create and manage multiple chats in IntelliJ Gemini Code Assist. Each chat contains its own context separate from other chats.