Adapters

Introduction

In DIAL, model adapters unify provider-specific model APIs with the Unified Protocol of DIAL Core. Each adapter consists of:

• Coded implementation that talks to the AI model and implements the Unified Protocol.
• Metadata object that you manage in Builders/Adapters, which establishes the relationship to the models.

DIAL includes adapters for Azure OpenAI models, GCP Vertex AI models, and AWS Bedrock models.

Compatibility with Azure OpenAI API, makes it simple to add new adapters for language models or develop them with DIAL SDK.

DIAL enables use of self-hosted (deployed withing the DIAL infrastructure) and external adapters.

Main Screen

The main screen displays all registered adapters in your DIAL instance.

Adapters grid

Column	Description
ID	Unique identifier of the adapter.
Display Name	Name of the adapter displayed on UI.
Description	Brief description of the adapter (e.g., "Adapter for OpenAI models").
Updated Time	Timestamp of the last update.
Creation Time	Creation timestamp.
Topics	Semantic tags associated with adapter.
Source type	Type of the adapter source: external endpoint or adapter container.
Source	URL for externally-deployed adapters or the name of the adapter container for self-hosted adapters.

Create

On the main screen, you can add new adapters by following these steps:

1. Click + Create to invoke the Create Adapter modal.

   Field	Required	Description
   ID	Yes	Unique identifier.
   Display name	Yes	Unique name of the adapter displayed on UI.
   Description	No	Description of the adapter.
   Source type	Yes	External Endpoint for externally-deployed adapters or Adapter Container for self-hosted adapter images.
   Completion endpoint	Yes	Endpoint that will be used to process chat completion requests. Implements the Unified Protocol (following the format: {ADAPTER_ORIGIN}/openai/deployments/). If Source Type is Adapter Container, the based URL is determined by the selected container and the endpoint path is editable. If Source Type is External Endpoint, the URL is fully editable.
   Responses endpoint	No	Endpoint of the model adapter that supports the OpenAI Responses API. Currently only OpenAI adapters support this. When set, DIAL Core routes POST /openai/v1/responses requests to this endpoint. Only basic Responses API behavior is supported: background requests, previous_request_id, conversations, prompts, and files are not supported. If Source Type is Adapter Container, the based URL is determined by the selected container and the endpoint path is editable. If Source Type is External Endpoint, the URL is fully editable.
   Container	Conditional	Name of the running adapter container. Click to select from the list of available containers. Applies to Adapter Container source type.

2. Once all required fields are filled, click Create. The dialog closes and the new adapter's configuration screen is opened. A new adapter will appear immediately on the main screen once it is created.

   

Configuration

Click any adapter on the main screen to open its configuration page.

Top Bar Controls

• Create Model: Use to create a model deployment using the selected model adapter as a source type. Created models will be available in the Entities/Models section.
• Delete: Use to remove the adapter itself and all models utilizing it. Note: After confirmation - the adapter and all related models are deleted.
• Save: Use to save and apply any changes.
• Discard: Use to discard any unsaved changes. You can revert changes in the Audit section.
• JSON Editor (Toggle): Switch between the form-based UI and raw JSON view of the adapter's configuration. Use JSON mode for copy-paste or advanced edits.

Properties

In the Properties tab, you can view and define identity and metadata of the selected adapter.

Field	Required	Editable	Description
ID	-	No	Unique read-only identifier of the adapter.
Updated Time	-	No	Timestamp of the last update to this adapter's configuration.
Creation Time	-	No	Adapter creation timestamp.
Display Name	Yes	Yes	Unique name of the adapter displayed on UI.
Description	No	Yes	Brief description of the adapter.
Source type	Yes	Yes	External Endpoint for externally-deployed adapters or Adapter Container for self-hosted adapter images.
Completion endpoint path	Yes	Yes	Endpoint URL of the adapter service that will be used to process chat completion requests. Implements the Unified Protocol (following the format: {ADAPTER_ORIGIN}/openai/deployments/). Responses API endpoint is recommended to use here for OpenAI models that support Responses API. If Source Type is Adapter Container, the base URL is determined by the selected container, while the path can be partially customized. If Source Type is External Endpoint, the URL is fully editable.
Responses endpoint path	No	Yes	Endpoint of the model adapter that supports the OpenAI Responses API. Currently only OpenAI adapters support this. When set, DIAL Core routes POST /openai/v1/responses requests to this endpoint. Only basic Responses API behavior is supported: background requests, previous_request_id, conversations, prompts, and files are not supported. If Source Type is Adapter Container, the base URL is determined by the selected Adapter container, while the path can be partially customized. If Source Type is External Endpoint, the URL is fully editable.
Container	Conditional	Yes	Name of the running Adapter Container. Click to select from the list of available containers. Applies and is required to Adapter Container source type.
Topics	No	Yes	Semantic tags associated with adapter. Click to display a list of available topics. You can add your own custom topics to the list following these rules: - The topic name must not exceed 255 characters. - The topic name must not contain leading or trailing spaces.

Models

In the Models tab, you can view and add AI models exposed by the selected adapter.

Field	Description
ID	Model's identifier.
Display Name	Name of the AI model displayed on UI.
Description	Description of the model.

Add

You can add AI models that will be processed by the selected adapter.

1. Click + Add.
2. Select one or more available models in the modal window. You can check all the available models in the Entities → Models section. You can also use + Create Model button on this screen to create a new model on the fly.
3. Click Apply to insert them into the table.

Remove

You can remove models processed by the adapter.

1. Click the actions menu in the model’s line.
2. Choose Remove in the menu.

Audit

On this screen, you can access a detailed preview and revert any changes made to the selected AI model adapter.

TIP: This section mimics the functionality available in the global Audit → Activities menu, but is scoped specifically to the selected AI model adapter.

JSON Editor

Advanced users with technical expertise can work with the adapter properties in a JSON editor view mode. It is useful for advanced scenarios of bulk updates, copy/paste between environments, or tweaking settings not exposed in the form UI.

Switching to the JSON Editor

1. Navigate to Builders → Adapters, then select the adapter you want to edit.
2. Click the JSON Editor toggle (top-right). The UI reveals the raw JSON.

TIP: Switching modes is disabled if there are any unsaved changes.