Schema for creating custom integrations.
Inherits api_key field validation from CustomConfig:
Token field is only used for oauth2 authentication (contains OAuth2 client credentials). For api_key auth, the api_key_value field is used instead.
Endpoint URL for the custom integration.
Configuration for multi-modal (file upload) capabilities.
Authentication types for custom integrations.
Values:
api_key, none, oauth2 List of model names for the custom integration. Deprecated: use model_properties instead.
List of model properties with name and alias for the custom integration.
Internal: whether this config was created from the legacy 'models' field.
Default model to use. If not provided, defaults to the first model.
Optional scope for OAuth2 authentication.
OAuth2 token URL for custom OAuth2 authentication. If not provided, defaults to the endpoint.
HTTP header name to use for API key authentication (e.g., 'X-API-Key', 'Authorization').
API key value to send in the specified header for authentication.
Optional configuration for a custom LiteLLM handler class. When specified, the handler's acompletion() method is used instead of the default litellm.acompletion().
Custom header mapping from internal fields (job_id, user_id, project_id, run_id) to custom header names to be included in LLM requests.
Optional custom HTTP headers to include in requests to the integration endpoint. Stored encrypted at rest.
Successful Response
anthropic, aws_bedrock, aws_sagemaker, azure, custom, databricks, labelstudio, mistral, nvidia, openai, vegas_gateway, vertex_ai, writer