AI Settings: Custom Models (Bring Your Own Key)

Bring Your Own Key (BYOK) lets your organization power Polaris and Custom Agents with your own provider instead of using the Contentstack Managed LLM. You configure your provider credentials from the Custom Models tab in AI Settings, and Contentstack sends the AI requests for that product to your provider account.

Note Bring Your Own Key must be enabled for your organization before you can configure a custom model. To request access, contact your Contentstack account manager.

Bring Your Own Key supports two products, Polaris and Custom Agents, and each supports its own set of providers:

• Polaris: OpenAI (GPT-5.2 and above).
• Custom Agent: OpenAI, Azure OpenAI, Google Vertex AI, Google Gemini, and Anthropic.

You configure each product's provider credentials from the Custom Models page. For Custom Agents, the provider you configure becomes the Organization Default model for your agents. You can also override the Organization Default model for an individual agent in its Settings.

• Contentstack account
• Organization Owner or Admin permissions
• Bring Your Own Key enabled for your organization
• Credentials from your LLM provider, such as an API key, or a service account JSON key for Google Vertex AI

• How to set up a product-level custom model configuration for Polaris and Custom Agent.
• How to set up a custom model configuration for an individual agent.
• How to switch a product between Managed by Contentstack and a custom configuration.
• Where custom model details appear in agent settings and the Executions tab.

You can configure a custom model at two levels:

• Product-level configuration: A custom model configured for a product that supports Bring Your Own Key (Polaris or Custom Agent), so that product uses your provider and credentials.
• Agent-level configuration: A custom model configured for an individual agent, set from the agent's settings.

By default, products use the Contentstack Managed LLM. When you configure and apply a custom model, the associated AI features run on your own provider account instead.

To manage custom models, log in to your Contentstack account and follow the steps:

1. Navigate to the App Switcher icon in the top-right corner and click Administration.
2. From the top header, select AI Settings.
3. In the left navigation, click Custom Models.

The Custom Models page lists the AI products that support Bring Your Own Key, Polaris and Custom Agent, each with its Configuration Type (for example, Managed by Contentstack), its connection Status, and the date it was Last Updated.

Click to enlarge

Configure a custom model for a product that supports Bring Your Own Key (Polaris or Custom Agent). The fields you enter depend on the provider. All fields are required.

Provider	Fields
OpenAI	Model ID (for example, gpt-4o) and API Key
Azure OpenAI	Deployment ID (the deployment name of your Azure OpenAI resource), Resource Name (the name of your Azure resource), API Key, and API Version
Google Vertex AI	Model ID (for example, gemini-2.5-pro) and Service Account JSON Key (paste the full Google Cloud service account JSON key)
Google Gemini	Model ID (for example, gemini-2.5-pro) and API Key
Anthropic	Model ID (for example, claude-sonnet-5) and API Key

In Model ID, enter the exact model identifier your provider expects, not the model's display name.

To set up a product-level configuration, perform the following steps:

1. On the Custom Models page, in the row for the product you want to configure, click the Actions (vertical ellipsis) icon and select Configure.
2. In the Configure LLM For [Product] panel, use the Choose Configuration drop-down to select how the product connects:
   1. Custom configuration: uses your own provider and credentials for this product.
   2. Managed by Contentstack: uses the Contentstack Managed LLM.
3. If you select Custom configuration, select the Provider under Custom Configuration Settings. Polaris offers OpenAI. Custom Agent offers OpenAI, Azure OpenAI, Google Vertex AI, Google Gemini, and Anthropic.
4. Enter the fields for the provider you selected, as listed in the table above.
5. Click Save Configuration.

Your credentials are stored securely and encrypted, and only the product you configure can use them.

Note To stop using your own credentials for a product, reopen the panel and set Choose Configuration back to Managed by Contentstack. The product then uses the Contentstack Managed LLM again.

Click to enlarge

You can configure a custom model for an individual agent, independent of the product configuration. When you add an API key for an agent, the Authorize window asks for a Title for the key and the provider's credentials. All fields are required.

Provider	Credentials
OpenAI	API Key and Organization ID
Azure OpenAI	Resource Name (the sub-domain in https://<resource-name>.openai.azure.com), API Key, Deployment Name/ID, and API Version
Google Vertex	Service Account Key
Gemini	API Key
Anthropic	API Key

To set up a custom model for an agent, perform the following steps:

1. Open Agent OS, go to Agents, and select your agent.
2. In the left navigation, click Settings.
3. In the AI Model section, click Change.
4. In the Change Model Setup panel, select Custom Configuration. The other option, Organization Default, keeps the agent on the model configured for the Custom Agent product in AI Settings.
5. Enter the following details:
   1. Select Model Provider (required): Select the LLM provider for this agent: OpenAI, Azure OpenAI, Google Vertex, Gemini, or Anthropic.
   2. Authentication Method (required): Select an existing API key, or click + Add API Key, enter the credentials in the Authorize window (see the table above), and click Authorize.
   3. Select AI Model (required): Select the model the agent should use.
6. Click Save.

Note If you don't set a custom model for an agent, the agent uses Organization Default, which follows the Custom Agent product configuration in AI Settings. For more about provider and model options at the agent level, refer to the Get Started with Agents documentation.

After you configure a custom model, Contentstack surfaces the model information in the following places:

• Agent settings: The AI Model section displays the provider and model, labeled Organization Custom Model or Custom Configuration, along with text indicating that this is your organization's own provider and key. When the agent uses the Contentstack Managed LLM, the section shows Contentstack Managed and doesn't expose the underlying model.
• Executions: In the agent's Executions tab, open a run and, in its Execution Steps, the Metrics section displays the Total Tokens used, the AI Provider, and the Model when a run uses a custom model. When a run uses the Contentstack Managed LLM, these custom model details are not shown.

Note When Polaris runs on a custom model, the Polaris panel does not display which model it is using. Agents display the model details in the agent settings and execution logs.

If your organization doesn't have Bring Your Own Key enabled, AI Settings doesn't show the Custom Models tab. In an agent's Change Model Setup panel, selecting Custom Configuration displays the following message:

Bring Your Own Key (BYOK) is not enabled for your organization. Please contact the Contentstack support team.

To enable Bring Your Own Key for your organization, contact your Contentstack account manager.

• To enable or disable AI features per product, refer to the AI Configuration : Global AI Settings documentation.
• To configure the AI model for an agent, refer to the Get Started with Agents documentation.
• To learn more about Polaris, refer to the Get Started with Polaris documentation.