Manage web app features

This document describes how to configure the feature management settings for your Gemini Enterprise web app using the Google Cloud console. You can control the features that are available for your end users.

Before you begin

Make sure you have the following:

Configure feature controls

To configure the features available in your web app, follow these steps:

  1. In the Google Cloud console, go to the Gemini Enterprise page.

    Gemini Enterprise

  2. Click the name of the app that you want to configure.

  3. Click Configurations and then click the Feature Management tab.

  4. You can turn on or turn off the following features in the web app:

    Feature control Description
    Enable Agent Gallery When enabled, the web app shows the Agent Gallery. For more information, see Browse agents with Agent Gallery.
    Enable chat agents When enabled, users can build and use custom chat agents. For more information, see Chat agent overview.
    Enable workflows When enabled, users can create and run custom workflows. For more information, see Workflows overview.
    Enable model selector When enabled, users can select the Gemini or partner model that they want to use in the web app.
    Enable Gemini Notebook When enabled, users can use Gemini Notebook Enterprise in the web app. For more information, see Use Gemini Notebook Enterprise within Gemini Enterprise.
    Enable session sharing When enabled, users can share Gemini Enterprise conversations with other users.
    Enable memory and customization When enabled, Gemini Enterprise uses past conversations and saved memory settings to generate personalized and relevant responses.
    Enable Canvas When enabled, users can create, edit, and export files in an interactive editor next to their chat conversation. For more information, see Create and edit documents and slides in Canvas.
    Enable projects When enabled, users can create and use collaborative projects in the web app. For more information, see Create and manage projects.
    Enable image generation When enabled, users can generate images in the web app. You can also choose the image model you want your users to use.

    Available models include:
    • Gemini 3 Pro image (Nano Banana Pro): This model is generally available (GA) and is turned off by default. It is only available in the Global region.
    • Gemini 3.1 Flash image (Nano Banana 2): This model is generally available (GA) and is turned off by default. If an administrator turned on this model during the Public Preview, it remains turned on in GA too.
    • Gemini 2.5 Flash image: The system defaults to this model if you don't make any selection.
    Enable video generation When enabled, users can generate videos in the web app.
    Enable OneDrive upload When enabled, users can upload files to the assistant from Microsoft OneDrive as a data source.
    Enable talk to content When enabled, users can navigate to the Sources referenced in an answer, add documents to the conversation by clicking the add icon next to a source, and then ask questions about the content of the documents.
    Enable speech-to-text When enabled, users can dictate their prompts into the web app, using the mic icon in the chat box. For more information, see Use your voice to chat.
    Enable Google Drive upload When enabled, users can upload files to the assistant from Google Drive as a data source.
    Include cross-domain documents You can enable the Include cross-domain documents feature for an existing app, even if you didn't enable it during app creation. When enabled, users can search and access documents from outside your organization using the Google Drive data store.

    Enable welcome emails When enabled, users receive a welcome email the first time they sign in to the web app.
    Enable skills When enabled, users can create and use skills in the web app. For more information, see Create and manage skills.
    Enable skill sharing When enabled, users can share custom skills with other users using the Gemini Enterprise app. To enable this, you must turn on Enable skills. If you don't also turn on the Enable skill sharing without admin approval setting, administrators must review and approve pending skill share requests. For more information, see Create and manage skills.
    Enable skill sharing without admin approval When enabled, users on your team can share and use custom skills without administrator approval when using the Gemini Enterprise app. To enable this setting, you must turn on Enable skills and Enable skill sharing. For more information, see Create and manage skills.
    Enable agent sharing When enabled, users can share agents created using Workflow Builder with other users using the Gemini Enterprise app. For more information, see Share an agent.

    If you use Okta or Microsoft Entra ID as a third-party identity provider with Workforce Identity Federation, configure SCIM to enable autocomplete of email addresses (and group names) when users share agents.

    Enable agent sharing without admin approval When enabled, users on your team can share and use agents created using Workflow Builder without administrator approval when using the Gemini Enterprise app.

    Note: Review and approve pending sharing requests before turning on this feature to prevent users from having to resubmit the sharing requests.

    For more information, see Share an agent.
    Enable End Users to share with Groups When enabled, users can share custom agents with groups in the organization. Whether the administrator must approve sharing depends on the Enable agent sharing without admin approval setting. For more information, see Share an agent.

    If you use Okta or Microsoft Entra ID as a third-party identity provider with Workforce Identity Federation, configure SCIM to enable autocomplete of group names (and email addresses) when users share agents.

    Model availability You can select the models that users can use in the web app. To enable model selection, you must turn on the Enable model selector toggle.

    The following Gemini and partner models are available:

    • Gemini models: Gemini 4 Argon, Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.1 Pro, and Gemini 2.5 Pro. For more information, see Enable Gemini models.
    • Partner models: Anthropic models—Claude Opus 5.5 and Claude Sonnet 5.5. For more information, see Enable third-party models.

Enable Gemini models

You can control which Gemini models users can select in the Gemini Enterprise web app. To let users choose a model in the web app, make sure that the Enable model selector toggle is turned on in the Feature Management tab.

Available Gemini models

The following Gemini models are available in the web app:

Model Default state Supported locations and notes
Gemini 4 Argon Turned off in global, us, and eu (defaults to on beginning October 27, 2026) To make Gemini 4 Argon available to users, click the Gemini 4 Argon toggle to the on position. For unsupported in-country regions, confirm the warning dialog to route traffic to the global endpoint. For more information, see Data residency for Gemini Enterprise.
Gemini 3.8 Flash Turned on in global, us, and eu, and in sg and europe-west2 If you don't want users to use Gemini 3.8 Flash, click the Gemini 3.8 Flash toggle to the off position. For unsupported in-country regions, click the Gemini 3.8 Flash toggle to the on position and confirm the warning dialog to route traffic to the global endpoint.
Gemini 3.7 Flash Turned off To let users use Gemini 3.7 Flash, click the Gemini 3.7 Flash toggle to the on position. For unsupported in-country regions, you must also confirm a warning dialog to route traffic to the global endpoint.
Gemini 3.6 Flash Turned off To let users use Gemini 3.6 Flash, click the Gemini 3.6 Flash toggle to the on position. For unsupported in-country regions, you must also confirm a warning dialog to route traffic to the global endpoint.
Gemini 3.5 Flash Turned on For some models in General Availability (GA) in the Gemini Enterprise app, such as Gemini 3.5 Flash, you can't turn the toggle off.
Gemini 3.1 Pro Turned off Available in Limited Availability if supported in your region. To let users use Gemini 3.1 Pro, click the Gemini 3.1 Pro toggle to the on position. For more information about known limitations, see Using Gemini 3.1 Pro in Limited Availability.
Gemini 2.5 Pro Turned on in global, us, and eu If you don't want users to use Gemini 2.5 Pro, click the Gemini 2.5 Pro toggle to the off position. Gemini 2.5 Pro is deprecated in the ca and asia-northeast1 regions and isn't available in those regions after October 20, 2026. For more information, see Data residency for Gemini Enterprise.

Turn on or turn off a Gemini model

To turn on or turn off a Gemini model in the web app, follow these steps:

  1. In the Google Cloud console, go to the Gemini Enterprise page.

    Gemini Enterprise

  2. Click the name of the app that you want to configure.

  3. Click Configurations and then click the Feature Management tab.

  4. Go to the Gemini model availability section.

  5. Click the toggle for the Gemini model that you want to turn on or turn off.

    If you turn on a model in an unsupported in-country region, confirm the warning dialog to route traffic to the global endpoint.

  6. Click Save.

The changes take effect after a few minutes.

Enable third-party models

In addition to Gemini models, you can let users chat with third-party models, such as Anthropic models—Claude Opus 5.5 and Claude Sonnet 5.5—in the Gemini Enterprise web app. You pay for third-party model usage per token, on a pay-as-you-go basis.

Available third-party models

The following third-party models are available in the web app:

Model Publisher Supported locations
Claude Opus 5.5 Anthropic global, us, and eu
Claude Sonnet 5.5 Anthropic global, us, and eu

Anthropic models—Claude Opus 5.5 and Claude Sonnet 5.5—are turned off by default in the global, us, and eu regions. To make them available to end users, turn them on by enabling billing and accepting the terms of service. For more information about regional support, see Data residency for Gemini Enterprise.

Third-party models aren't available in the Gemini Enterprise Starter edition. They also aren't available for in-country regions.

Before you begin

Before you turn on a third-party model, make sure that the following requirements are met:

  • You have the Gemini Enterprise Admin IAM role (roles/discoveryengine.agentspaceAdmin).

  • If you have a subscription, Pay-as-you-go usage above quota is turned on for your project. You must turn it on separately for each edition, such as Gemini Enterprise Standard and Gemini Enterprise Plus. For more information, see Configure overages.

  • The Enable model selector toggle is turned on in the Feature Management tab, so that users can select the model in the web app.

We recommend that you set a monthly spend limit and budgets in Cloud Billing to control costs. For pricing information, see Claude pricing.

Turn on a third-party model

To make a third-party model available to users in the web app, follow these steps:

  1. In the Google Cloud console, go to the Gemini Enterprise page.

    Gemini Enterprise

  2. Click the name of the app that you want to configure.

  3. Click Configurations and then click the Feature Management tab.

  4. Go to the Enable third-party models section.

  5. In the section for the publisher, click the toggle for the model that you want to turn on.

    The Model pricing dialog appears. The dialog describes how the model is charged and checks whether Pay-as-you-go usage above quota is turned on for your project.

  6. In the Model pricing dialog, do one of the following:

    • If pay-as-you-go usage is available for your project, click Enable.
    • If Pay-as-you-go usage above quota isn't turned on, click Go to Usage & spend to turn it on, and then return to this procedure. The model toggle is turned off.
    • If your project's billing account isn't eligible for Pay-as-you-go usage above quota, you can't turn on the model. Click Cancel and contact your billing administrator.
  7. If the Terms of service dialog appears, read the terms for the model. Go to the end of the terms, and then click Agree.

    You accept the terms of service once for each project and model family. For example, after you accept the terms for Claude Opus 5.5 in a project, the Terms of service dialog doesn't appear again when you turn on that model in other apps in the same project.

    If you click Cancel, the model toggle is turned off.

  8. Click Save.

  9. In the Important billing info dialog, review the billing information, and then click I understand.

Turn off a third-party model

To stop users from using a third-party model, follow these steps:

  1. In the Feature Management tab, go to the Enable third-party models section.

  2. Click the toggle for the model to the off position.

  3. Click Save.

The model is removed from the model selector in the web app after a few minutes.