Chat Service

Specifies the unique identifier for the model exposed by this deployment.

• For OpenAI, this is typically a string such as gpt-4o or text-embedding-ada-002.
• For Azure OpenAI, use the deployment name or model ID as configured in the Azure portal.

This value is used to select the deployment for LLM operations within Grooper.

The full endpoint URL for the API service.

• For OpenAI, use https://api.openai.com/v1/...
• For Azure OpenAI, use the endpoint provided in the Azure portal (e.g., https://your-resource.openai.azure.com/openai/deployments/your-deployment/...)

Ensure this URL matches the requirements of your service provider and deployment.

Can be one of the following values:

Selects the authentication scheme required by your API endpoint.

Specifies the secret key or token required to authenticate API requests for this deployment.

• The value entered here is used according to the selected 'Authorization' method.
  • For 'Bearer', this is typically an OpenAI API key (e.g., sk-...).
  • For 'API Key', use the key provided by your service (such as Azure OpenAI).
  • Leave blank if 'Authorization' is set to 'None'.
• This value is sent with every API request as an HTTP header, either as Authorization: Bearer <APIKey> or api-key: <APIKey>, depending on the method.

Security Note: The API key is sensitive and should be protected. Do not share or expose this value in public repositories, logs, or user interfaces.

Allows you to specify additional HTTP headers to include with every API request.

• Enter each header as a key=value pair, one per line.
• Example:
• These headers are merged with the default authorization headers. If a custom header matches a default header (such as Authorization or api-key), your value will override the default.
• Use this property to provide any extra metadata or configuration required by your API provider, such as organization IDs, region, or custom authentication tokens.
• Leave blank if no additional headers are required.

Enabling this property causes contextual information to be submitted as "system" messages rather than "user" messages. This can help the LLM distinguish which parts of the conversation are instructions and which parts are contextual information.

System messages are supported on most OpenAI-trained models, but may not work properly with models trained by other organizations.