diff --git a/acceptance/bundle/refschema/out.fields.txt b/acceptance/bundle/refschema/out.fields.txt
index db4fbe39c1..d950ba064a 100644
--- a/acceptance/bundle/refschema/out.fields.txt
+++ b/acceptance/bundle/refschema/out.fields.txt
@@ -2252,6 +2252,475 @@ resources.jobs.*.permissions.permissions[*].group_name	string	ALL
 resources.jobs.*.permissions.permissions[*].permission_level	iam.PermissionLevel	ALL
 resources.jobs.*.permissions.permissions[*].service_principal_name	string	ALL
 resources.jobs.*.permissions.permissions[*].user_name	string	ALL
+resources.model_serving_endpoints.*.ai_gateway	*serving.AiGatewayConfig	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.fallback_config	*serving.FallbackConfig	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.fallback_config.enabled	bool	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails	*serving.AiGatewayGuardrails	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.input	*serving.AiGatewayGuardrailParameters	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.input.invalid_keywords	[]string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.input.invalid_keywords[*]	string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.input.pii	*serving.AiGatewayGuardrailPiiBehavior	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.input.pii.behavior	serving.AiGatewayGuardrailPiiBehaviorBehavior	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.input.safety	bool	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.input.valid_topics	[]string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.input.valid_topics[*]	string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.output	*serving.AiGatewayGuardrailParameters	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.output.invalid_keywords	[]string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.output.invalid_keywords[*]	string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.output.pii	*serving.AiGatewayGuardrailPiiBehavior	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.output.pii.behavior	serving.AiGatewayGuardrailPiiBehaviorBehavior	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.output.safety	bool	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.output.valid_topics	[]string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.guardrails.output.valid_topics[*]	string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.inference_table_config	*serving.AiGatewayInferenceTableConfig	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.inference_table_config.catalog_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.inference_table_config.enabled	bool	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.inference_table_config.schema_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.inference_table_config.table_name_prefix	string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.rate_limits	[]serving.AiGatewayRateLimit	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.rate_limits[*]	serving.AiGatewayRateLimit	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.rate_limits[*].calls	int64	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.rate_limits[*].key	serving.AiGatewayRateLimitKey	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.rate_limits[*].principal	string	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.rate_limits[*].renewal_period	serving.AiGatewayRateLimitRenewalPeriod	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.rate_limits[*].tokens	int64	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.usage_tracking_config	*serving.AiGatewayUsageTrackingConfig	INPUT	STATE
+resources.model_serving_endpoints.*.ai_gateway.usage_tracking_config.enabled	bool	INPUT	STATE
+resources.model_serving_endpoints.*.budget_policy_id	string	INPUT	STATE
+resources.model_serving_endpoints.*.config	*serving.EndpointCoreConfigInput	INPUT	STATE
+resources.model_serving_endpoints.*.config.auto_capture_config	*serving.AutoCaptureConfigInput	INPUT	STATE
+resources.model_serving_endpoints.*.config.auto_capture_config.catalog_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.auto_capture_config.enabled	bool	INPUT	STATE
+resources.model_serving_endpoints.*.config.auto_capture_config.schema_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.auto_capture_config.table_name_prefix	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities	[]serving.ServedEntityInput	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*]	serving.ServedEntityInput	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].entity_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].entity_version	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].environment_vars	map[string]string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].environment_vars.*	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model	*serving.ExternalModel	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.ai21labs_config	*serving.Ai21LabsConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.ai21labs_config.ai21labs_api_key	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.ai21labs_config.ai21labs_api_key_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.amazon_bedrock_config	*serving.AmazonBedrockConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.amazon_bedrock_config.aws_access_key_id	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.amazon_bedrock_config.aws_access_key_id_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.amazon_bedrock_config.aws_region	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.amazon_bedrock_config.aws_secret_access_key	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.amazon_bedrock_config.aws_secret_access_key_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.amazon_bedrock_config.bedrock_provider	serving.AmazonBedrockConfigBedrockProvider	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.amazon_bedrock_config.instance_profile_arn	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.anthropic_config	*serving.AnthropicConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.anthropic_config.anthropic_api_key	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.anthropic_config.anthropic_api_key_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.cohere_config	*serving.CohereConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.cohere_config.cohere_api_base	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.cohere_config.cohere_api_key	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.cohere_config.cohere_api_key_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config	*serving.CustomProviderConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config.api_key_auth	*serving.ApiKeyAuth	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config.api_key_auth.key	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config.api_key_auth.value	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config.api_key_auth.value_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config.bearer_token_auth	*serving.BearerTokenAuth	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config.bearer_token_auth.token	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config.bearer_token_auth.token_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.custom_provider_config.custom_provider_url	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.databricks_model_serving_config	*serving.DatabricksModelServingConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.databricks_model_serving_config.databricks_api_token	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.databricks_model_serving_config.databricks_api_token_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.databricks_model_serving_config.databricks_workspace_url	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.google_cloud_vertex_ai_config	*serving.GoogleCloudVertexAiConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.google_cloud_vertex_ai_config.private_key	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.google_cloud_vertex_ai_config.private_key_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.google_cloud_vertex_ai_config.project_id	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.google_cloud_vertex_ai_config.region	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config	*serving.OpenAiConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.microsoft_entra_client_id	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.microsoft_entra_client_secret	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.microsoft_entra_client_secret_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.microsoft_entra_tenant_id	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.openai_api_base	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.openai_api_key	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.openai_api_key_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.openai_api_type	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.openai_api_version	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.openai_deployment_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.openai_config.openai_organization	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.palm_config	*serving.PaLmConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.palm_config.palm_api_key	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.palm_config.palm_api_key_plaintext	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.provider	serving.ExternalModelProvider	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].external_model.task	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].instance_profile_arn	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].max_provisioned_concurrency	int	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].max_provisioned_throughput	int	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].min_provisioned_concurrency	int	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].min_provisioned_throughput	int	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].provisioned_model_units	int64	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].scale_to_zero_enabled	bool	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].workload_size	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_entities[*].workload_type	serving.ServingModelWorkloadType	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models	[]serving.ServedModelInput	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*]	serving.ServedModelInput	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].environment_vars	map[string]string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].environment_vars.*	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].instance_profile_arn	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].max_provisioned_concurrency	int	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].max_provisioned_throughput	int	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].min_provisioned_concurrency	int	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].min_provisioned_throughput	int	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].model_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].model_version	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].provisioned_model_units	int64	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].scale_to_zero_enabled	bool	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].workload_size	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.served_models[*].workload_type	serving.ServedModelInputWorkloadType	INPUT	STATE
+resources.model_serving_endpoints.*.config.traffic_config	*serving.TrafficConfig	INPUT	STATE
+resources.model_serving_endpoints.*.config.traffic_config.routes	[]serving.Route	INPUT	STATE
+resources.model_serving_endpoints.*.config.traffic_config.routes[*]	serving.Route	INPUT	STATE
+resources.model_serving_endpoints.*.config.traffic_config.routes[*].served_entity_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.traffic_config.routes[*].served_model_name	string	INPUT	STATE
+resources.model_serving_endpoints.*.config.traffic_config.routes[*].traffic_percentage	int	INPUT	STATE
+resources.model_serving_endpoints.*.description	string	INPUT	STATE
+resources.model_serving_endpoints.*.email_notifications	*serving.EmailNotifications	INPUT	STATE
+resources.model_serving_endpoints.*.email_notifications.on_update_failure	[]string	INPUT	STATE
+resources.model_serving_endpoints.*.email_notifications.on_update_failure[*]	string	INPUT	STATE
+resources.model_serving_endpoints.*.email_notifications.on_update_success	[]string	INPUT	STATE
+resources.model_serving_endpoints.*.email_notifications.on_update_success[*]	string	INPUT	STATE
+resources.model_serving_endpoints.*.endpoint_details	*serving.ServingEndpointDetailed	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway	*serving.AiGatewayConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.fallback_config	*serving.FallbackConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.fallback_config.enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails	*serving.AiGatewayGuardrails	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.input	*serving.AiGatewayGuardrailParameters	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.input.invalid_keywords	[]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.input.invalid_keywords[*]	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.input.pii	*serving.AiGatewayGuardrailPiiBehavior	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.input.pii.behavior	serving.AiGatewayGuardrailPiiBehaviorBehavior	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.input.safety	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.input.valid_topics	[]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.input.valid_topics[*]	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.output	*serving.AiGatewayGuardrailParameters	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.output.invalid_keywords	[]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.output.invalid_keywords[*]	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.output.pii	*serving.AiGatewayGuardrailPiiBehavior	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.output.pii.behavior	serving.AiGatewayGuardrailPiiBehaviorBehavior	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.output.safety	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.output.valid_topics	[]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.guardrails.output.valid_topics[*]	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.inference_table_config	*serving.AiGatewayInferenceTableConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.inference_table_config.catalog_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.inference_table_config.enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.inference_table_config.schema_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.inference_table_config.table_name_prefix	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.rate_limits	[]serving.AiGatewayRateLimit	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.rate_limits[*]	serving.AiGatewayRateLimit	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.rate_limits[*].calls	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.rate_limits[*].key	serving.AiGatewayRateLimitKey	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.rate_limits[*].principal	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.rate_limits[*].renewal_period	serving.AiGatewayRateLimitRenewalPeriod	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.rate_limits[*].tokens	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.usage_tracking_config	*serving.AiGatewayUsageTrackingConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.ai_gateway.usage_tracking_config.enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.budget_policy_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config	*serving.EndpointCoreConfigOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config	*serving.AutoCaptureConfigOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.catalog_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.schema_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.state	*serving.AutoCaptureState	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.state.payload_table	*serving.PayloadTable	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.state.payload_table.name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.state.payload_table.status	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.state.payload_table.status_message	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.auto_capture_config.table_name_prefix	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.config_version	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities	[]serving.ServedEntityOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*]	serving.ServedEntityOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].creation_timestamp	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].creator	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].entity_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].entity_version	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].environment_vars	map[string]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].environment_vars.*	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model	*serving.ExternalModel	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.ai21labs_config	*serving.Ai21LabsConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.ai21labs_config.ai21labs_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.ai21labs_config.ai21labs_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.amazon_bedrock_config	*serving.AmazonBedrockConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.amazon_bedrock_config.aws_access_key_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.amazon_bedrock_config.aws_access_key_id_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.amazon_bedrock_config.aws_region	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.amazon_bedrock_config.aws_secret_access_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.amazon_bedrock_config.aws_secret_access_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.amazon_bedrock_config.bedrock_provider	serving.AmazonBedrockConfigBedrockProvider	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.amazon_bedrock_config.instance_profile_arn	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.anthropic_config	*serving.AnthropicConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.anthropic_config.anthropic_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.anthropic_config.anthropic_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.cohere_config	*serving.CohereConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.cohere_config.cohere_api_base	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.cohere_config.cohere_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.cohere_config.cohere_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config	*serving.CustomProviderConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config.api_key_auth	*serving.ApiKeyAuth	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config.api_key_auth.key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config.api_key_auth.value	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config.api_key_auth.value_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config.bearer_token_auth	*serving.BearerTokenAuth	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config.bearer_token_auth.token	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config.bearer_token_auth.token_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.custom_provider_config.custom_provider_url	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.databricks_model_serving_config	*serving.DatabricksModelServingConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.databricks_model_serving_config.databricks_api_token	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.databricks_model_serving_config.databricks_api_token_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.databricks_model_serving_config.databricks_workspace_url	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.google_cloud_vertex_ai_config	*serving.GoogleCloudVertexAiConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.google_cloud_vertex_ai_config.private_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.google_cloud_vertex_ai_config.private_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.google_cloud_vertex_ai_config.project_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.google_cloud_vertex_ai_config.region	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config	*serving.OpenAiConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.microsoft_entra_client_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.microsoft_entra_client_secret	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.microsoft_entra_client_secret_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.microsoft_entra_tenant_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.openai_api_base	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.openai_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.openai_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.openai_api_type	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.openai_api_version	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.openai_deployment_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.openai_config.openai_organization	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.palm_config	*serving.PaLmConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.palm_config.palm_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.palm_config.palm_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.provider	serving.ExternalModelProvider	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].external_model.task	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].foundation_model	*serving.FoundationModel	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].foundation_model.description	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].foundation_model.display_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].foundation_model.docs	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].foundation_model.name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].instance_profile_arn	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].max_provisioned_concurrency	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].max_provisioned_throughput	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].min_provisioned_concurrency	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].min_provisioned_throughput	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].provisioned_model_units	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].scale_to_zero_enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].state	*serving.ServedModelState	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].state.deployment	serving.ServedModelStateDeployment	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].state.deployment_state_message	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].workload_size	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_entities[*].workload_type	serving.ServingModelWorkloadType	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models	[]serving.ServedModelOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*]	serving.ServedModelOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].creation_timestamp	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].creator	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].environment_vars	map[string]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].environment_vars.*	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].instance_profile_arn	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].max_provisioned_concurrency	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].min_provisioned_concurrency	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].model_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].model_version	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].provisioned_model_units	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].scale_to_zero_enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].state	*serving.ServedModelState	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].state.deployment	serving.ServedModelStateDeployment	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].state.deployment_state_message	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].workload_size	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.served_models[*].workload_type	serving.ServingModelWorkloadType	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.traffic_config	*serving.TrafficConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.traffic_config.routes	[]serving.Route	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.traffic_config.routes[*]	serving.Route	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.traffic_config.routes[*].served_entity_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.traffic_config.routes[*].served_model_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.config.traffic_config.routes[*].traffic_percentage	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.creation_timestamp	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.creator	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.data_plane_info	*serving.ModelDataPlaneInfo	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.data_plane_info.query_info	*serving.DataPlaneInfo	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.data_plane_info.query_info.authorization_details	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.data_plane_info.query_info.endpoint_url	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.description	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.email_notifications	*serving.EmailNotifications	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.email_notifications.on_update_failure	[]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.email_notifications.on_update_failure[*]	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.email_notifications.on_update_success	[]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.email_notifications.on_update_success[*]	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.endpoint_url	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.last_updated_timestamp	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config	*serving.EndpointPendingConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config	*serving.AutoCaptureConfigOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.catalog_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.schema_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.state	*serving.AutoCaptureState	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.state.payload_table	*serving.PayloadTable	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.state.payload_table.name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.state.payload_table.status	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.state.payload_table.status_message	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.auto_capture_config.table_name_prefix	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.config_version	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities	[]serving.ServedEntityOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*]	serving.ServedEntityOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].creation_timestamp	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].creator	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].entity_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].entity_version	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].environment_vars	map[string]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].environment_vars.*	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model	*serving.ExternalModel	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.ai21labs_config	*serving.Ai21LabsConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.ai21labs_config.ai21labs_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.ai21labs_config.ai21labs_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.amazon_bedrock_config	*serving.AmazonBedrockConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.amazon_bedrock_config.aws_access_key_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.amazon_bedrock_config.aws_access_key_id_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.amazon_bedrock_config.aws_region	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.amazon_bedrock_config.aws_secret_access_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.amazon_bedrock_config.aws_secret_access_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.amazon_bedrock_config.bedrock_provider	serving.AmazonBedrockConfigBedrockProvider	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.amazon_bedrock_config.instance_profile_arn	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.anthropic_config	*serving.AnthropicConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.anthropic_config.anthropic_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.anthropic_config.anthropic_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.cohere_config	*serving.CohereConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.cohere_config.cohere_api_base	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.cohere_config.cohere_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.cohere_config.cohere_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config	*serving.CustomProviderConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config.api_key_auth	*serving.ApiKeyAuth	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config.api_key_auth.key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config.api_key_auth.value	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config.api_key_auth.value_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config.bearer_token_auth	*serving.BearerTokenAuth	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config.bearer_token_auth.token	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config.bearer_token_auth.token_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.custom_provider_config.custom_provider_url	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.databricks_model_serving_config	*serving.DatabricksModelServingConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.databricks_model_serving_config.databricks_api_token	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.databricks_model_serving_config.databricks_api_token_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.databricks_model_serving_config.databricks_workspace_url	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.google_cloud_vertex_ai_config	*serving.GoogleCloudVertexAiConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.google_cloud_vertex_ai_config.private_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.google_cloud_vertex_ai_config.private_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.google_cloud_vertex_ai_config.project_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.google_cloud_vertex_ai_config.region	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config	*serving.OpenAiConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.microsoft_entra_client_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.microsoft_entra_client_secret	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.microsoft_entra_client_secret_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.microsoft_entra_tenant_id	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.openai_api_base	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.openai_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.openai_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.openai_api_type	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.openai_api_version	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.openai_deployment_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.openai_config.openai_organization	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.palm_config	*serving.PaLmConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.palm_config.palm_api_key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.palm_config.palm_api_key_plaintext	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.provider	serving.ExternalModelProvider	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].external_model.task	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].foundation_model	*serving.FoundationModel	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].foundation_model.description	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].foundation_model.display_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].foundation_model.docs	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].foundation_model.name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].instance_profile_arn	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].max_provisioned_concurrency	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].max_provisioned_throughput	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].min_provisioned_concurrency	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].min_provisioned_throughput	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].provisioned_model_units	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].scale_to_zero_enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].state	*serving.ServedModelState	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].state.deployment	serving.ServedModelStateDeployment	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].state.deployment_state_message	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].workload_size	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_entities[*].workload_type	serving.ServingModelWorkloadType	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models	[]serving.ServedModelOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*]	serving.ServedModelOutput	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].creation_timestamp	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].creator	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].environment_vars	map[string]string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].environment_vars.*	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].instance_profile_arn	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].max_provisioned_concurrency	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].min_provisioned_concurrency	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].model_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].model_version	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].provisioned_model_units	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].scale_to_zero_enabled	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].state	*serving.ServedModelState	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].state.deployment	serving.ServedModelStateDeployment	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].state.deployment_state_message	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].workload_size	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.served_models[*].workload_type	serving.ServingModelWorkloadType	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.start_time	int64	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.traffic_config	*serving.TrafficConfig	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.traffic_config.routes	[]serving.Route	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.traffic_config.routes[*]	serving.Route	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.traffic_config.routes[*].served_entity_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.traffic_config.routes[*].served_model_name	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.pending_config.traffic_config.routes[*].traffic_percentage	int	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.permission_level	serving.ServingEndpointDetailedPermissionLevel	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.route_optimized	bool	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.state	*serving.EndpointState	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.state.config_update	serving.EndpointStateConfigUpdate	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.state.ready	serving.EndpointStateReady	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.tags	[]serving.EndpointTag	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.tags[*]	serving.EndpointTag	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.tags[*].key	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.tags[*].value	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_details.task	string	REMOTE
+resources.model_serving_endpoints.*.endpoint_id	string	REMOTE
+resources.model_serving_endpoints.*.id	string	INPUT
+resources.model_serving_endpoints.*.lifecycle	resources.Lifecycle	INPUT
+resources.model_serving_endpoints.*.lifecycle.prevent_destroy	bool	INPUT
+resources.model_serving_endpoints.*.modified_status	string	INPUT
+resources.model_serving_endpoints.*.name	string	INPUT	STATE
+resources.model_serving_endpoints.*.permissions	[]resources.ModelServingEndpointPermission	INPUT
+resources.model_serving_endpoints.*.permissions[*]	resources.ModelServingEndpointPermission	INPUT
+resources.model_serving_endpoints.*.permissions[*].group_name	string	INPUT
+resources.model_serving_endpoints.*.permissions[*].level	resources.ModelServingEndpointPermissionLevel	INPUT
+resources.model_serving_endpoints.*.permissions[*].service_principal_name	string	INPUT
+resources.model_serving_endpoints.*.permissions[*].user_name	string	INPUT
+resources.model_serving_endpoints.*.rate_limits	[]serving.RateLimit	INPUT	STATE
+resources.model_serving_endpoints.*.rate_limits[*]	serving.RateLimit	INPUT	STATE
+resources.model_serving_endpoints.*.rate_limits[*].calls	int64	INPUT	STATE
+resources.model_serving_endpoints.*.rate_limits[*].key	serving.RateLimitKey	INPUT	STATE
+resources.model_serving_endpoints.*.rate_limits[*].renewal_period	serving.RateLimitRenewalPeriod	INPUT	STATE
+resources.model_serving_endpoints.*.route_optimized	bool	INPUT	STATE
+resources.model_serving_endpoints.*.tags	[]serving.EndpointTag	INPUT	STATE
+resources.model_serving_endpoints.*.tags[*]	serving.EndpointTag	INPUT	STATE
+resources.model_serving_endpoints.*.tags[*].key	string	INPUT	STATE
+resources.model_serving_endpoints.*.tags[*].value	string	INPUT	STATE
+resources.model_serving_endpoints.*.url	string	INPUT
+resources.model_serving_endpoints.*.permissions.object_id	string	ALL
+resources.model_serving_endpoints.*.permissions.permissions	[]iam.AccessControlRequest	ALL
+resources.model_serving_endpoints.*.permissions.permissions[*]	iam.AccessControlRequest	ALL
+resources.model_serving_endpoints.*.permissions.permissions[*].group_name	string	ALL
+resources.model_serving_endpoints.*.permissions.permissions[*].permission_level	iam.PermissionLevel	ALL
+resources.model_serving_endpoints.*.permissions.permissions[*].service_principal_name	string	ALL
+resources.model_serving_endpoints.*.permissions.permissions[*].user_name	string	ALL
 resources.models.*.creation_timestamp	int64	REMOTE
 resources.models.*.description	string	ALL
 resources.models.*.id	string	INPUT	REMOTE
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/databricks.yml.tmpl b/acceptance/bundle/resources/model_serving_endpoints/basic/databricks.yml.tmpl
new file mode 100644
index 0000000000..52269097f4
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/databricks.yml.tmpl
@@ -0,0 +1,11 @@
+bundle:
+  name: acc-$UNIQUE_NAME
+
+resources:
+  model_serving_endpoints:
+    my_endpoint:
+      name: $ENDPOINT_NAME
+
+      permissions:
+        - level: CAN_VIEW
+          user_name: deco-test-user@databricks.com
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-plan.direct.json
new file mode 100644
index 0000000000..8dbb050838
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-plan.direct.json
@@ -0,0 +1,39 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.my_endpoint": {
+      "action": "create",
+      "new_state": {
+        "config": {
+          "name": "[ENDPOINT_NAME_1]"
+        }
+      }
+    },
+    "resources.model_serving_endpoints.my_endpoint.permissions": {
+      "depends_on": [
+        {
+          "node": "resources.model_serving_endpoints.my_endpoint",
+          "label": "${resources.model_serving_endpoints.my_endpoint.endpoint_id}"
+        }
+      ],
+      "action": "create",
+      "new_state": {
+        "config": {
+          "object_id": "",
+          "permissions": [
+            {
+              "permission_level": "CAN_VIEW",
+              "user_name": "deco-test-user@databricks.com"
+            },
+            {
+              "permission_level": "CAN_MANAGE",
+              "service_principal_name": "[USERNAME]"
+            }
+          ]
+        },
+        "vars": {
+          "object_id": "/serving-endpoints/${resources.model_serving_endpoints.my_endpoint.endpoint_id}"
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-plan.terraform.json
new file mode 100644
index 0000000000..1d92fcbea4
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.my_endpoint": {
+      "action": "create"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-requests.direct.json b/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-requests.direct.json
new file mode 100644
index 0000000000..7231a1d7fc
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-requests.direct.json
@@ -0,0 +1,23 @@
+{
+  "body": {
+    "name": "[ENDPOINT_NAME_1]"
+  },
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints"
+}
+{
+  "body": {
+    "access_control_list": [
+      {
+        "permission_level": "CAN_VIEW",
+        "user_name": "deco-test-user@databricks.com"
+      },
+      {
+        "permission_level": "CAN_MANAGE",
+        "service_principal_name": "[USERNAME]"
+      }
+    ]
+  },
+  "method": "PUT",
+  "path": "/api/2.0/permissions/serving-endpoints/[ENDPOINT_ID_1]"
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-requests.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-requests.terraform.json
new file mode 100644
index 0000000000..70a6d2b840
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.first-requests.terraform.json
@@ -0,0 +1,23 @@
+{
+  "body": {
+    "name": "[ENDPOINT_NAME_1]"
+  },
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints"
+}
+{
+  "body": {
+    "access_control_list": [
+      {
+        "permission_level": "CAN_MANAGE",
+        "service_principal_name": "[USERNAME]"
+      },
+      {
+        "permission_level": "CAN_VIEW",
+        "user_name": "deco-test-user@databricks.com"
+      }
+    ]
+  },
+  "method": "PUT",
+  "path": "/api/2.0/permissions/serving-endpoints/[ENDPOINT_ID_1]"
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-plan.direct.json
new file mode 100644
index 0000000000..768c8eaac2
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-plan.direct.json
@@ -0,0 +1,66 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.my_endpoint": {
+      "action": "recreate",
+      "new_state": {
+        "config": {
+          "name": "[ENDPOINT_NAME_2]"
+        }
+      },
+      "changes": {
+        "local": {
+          "name": {
+            "action": "recreate"
+          }
+        }
+      }
+    },
+    "resources.model_serving_endpoints.my_endpoint.permissions": {
+      "depends_on": [
+        {
+          "node": "resources.model_serving_endpoints.my_endpoint",
+          "label": "${resources.model_serving_endpoints.my_endpoint.endpoint_id}"
+        }
+      ],
+      "action": "update",
+      "new_state": {
+        "config": {
+          "object_id": "",
+          "permissions": [
+            {
+              "permission_level": "CAN_VIEW",
+              "user_name": "deco-test-user@databricks.com"
+            },
+            {
+              "permission_level": "CAN_MANAGE",
+              "service_principal_name": "[USERNAME]"
+            }
+          ]
+        },
+        "vars": {
+          "object_id": "/serving-endpoints/${resources.model_serving_endpoints.my_endpoint.endpoint_id}"
+        }
+      },
+      "remote_state": {
+        "object_id": "/serving-endpoints/[ENDPOINT_ID_1]",
+        "permissions": [
+          {
+            "permission_level": "CAN_VIEW",
+            "user_name": "deco-test-user@databricks.com"
+          },
+          {
+            "permission_level": "CAN_MANAGE",
+            "service_principal_name": "[USERNAME]"
+          }
+        ]
+      },
+      "changes": {
+        "local": {
+          "object_id": {
+            "action": "update"
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-plan.terraform.json
new file mode 100644
index 0000000000..0e6bc43fd9
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.my_endpoint": {
+      "action": "recreate"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-requests.direct.json b/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-requests.direct.json
new file mode 100644
index 0000000000..6826868502
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-requests.direct.json
@@ -0,0 +1,27 @@
+{
+  "method": "DELETE",
+  "path": "/api/2.0/serving-endpoints/[ENDPOINT_NAME_1]"
+}
+{
+  "body": {
+    "name": "[ENDPOINT_NAME_2]"
+  },
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints"
+}
+{
+  "body": {
+    "access_control_list": [
+      {
+        "permission_level": "CAN_VIEW",
+        "user_name": "deco-test-user@databricks.com"
+      },
+      {
+        "permission_level": "CAN_MANAGE",
+        "service_principal_name": "[USERNAME]"
+      }
+    ]
+  },
+  "method": "PUT",
+  "path": "/api/2.0/permissions/serving-endpoints/[ENDPOINT_ID_2]"
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-requests.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-requests.terraform.json
new file mode 100644
index 0000000000..11bb519e1c
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.second-requests.terraform.json
@@ -0,0 +1,39 @@
+{
+  "body": {
+    "access_control_list": [
+      {
+        "permission_level": "CAN_MANAGE",
+        "user_name": "[USERNAME]"
+      }
+    ]
+  },
+  "method": "PUT",
+  "path": "/api/2.0/permissions/serving-endpoints/[ENDPOINT_ID_1]"
+}
+{
+  "method": "DELETE",
+  "path": "/api/2.0/serving-endpoints/[ENDPOINT_NAME_1]"
+}
+{
+  "body": {
+    "name": "[ENDPOINT_NAME_2]"
+  },
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints"
+}
+{
+  "body": {
+    "access_control_list": [
+      {
+        "permission_level": "CAN_MANAGE",
+        "service_principal_name": "[USERNAME]"
+      },
+      {
+        "permission_level": "CAN_VIEW",
+        "user_name": "deco-test-user@databricks.com"
+      }
+    ]
+  },
+  "method": "PUT",
+  "path": "/api/2.0/permissions/serving-endpoints/[ENDPOINT_ID_2]"
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/out.test.toml b/acceptance/bundle/resources/model_serving_endpoints/basic/out.test.toml
new file mode 100644
index 0000000000..f474b1b917
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/out.test.toml
@@ -0,0 +1,5 @@
+Local = false
+Cloud = true
+
+[EnvMatrix]
+  DATABRICKS_BUNDLE_ENGINE = ["terraform", "direct"]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/output.txt b/acceptance/bundle/resources/model_serving_endpoints/basic/output.txt
new file mode 100644
index 0000000000..b5400e5285
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/output.txt
@@ -0,0 +1,41 @@
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/acc-[UNIQUE_NAME]/default/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+
+=== Print the GET endpoint details
+{
+  "name": "[ENDPOINT_NAME_1]",
+  "creator": "[USERNAME]"
+}
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/acc-[UNIQUE_NAME]/default/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+
+=== Print the GET endpoint details
+{
+  "name": "[ENDPOINT_NAME_2]",
+  "creator": "[USERNAME]"
+}
+
+>>> [CLI] bundle destroy --auto-approve
+The following resources will be deleted:
+  delete model_serving_endpoint my_endpoint
+
+All files and directories at the following location will be deleted: /Workspace/Users/[USERNAME]/.bundle/acc-[UNIQUE_NAME]/default
+
+Deleting files...
+Destroy complete!
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/script b/acceptance/bundle/resources/model_serving_endpoints/basic/script
new file mode 100755
index 0000000000..5201fbeee2
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/script
@@ -0,0 +1,39 @@
+export ENDPOINT_NAME="test-endpoint-$UNIQUE_NAME-1"
+
+envsubst < databricks.yml.tmpl > databricks.yml
+
+cleanup() {
+    trace $CLI bundle destroy --auto-approve
+    rm out.requests.txt
+}
+trap cleanup EXIT
+
+trace $CLI bundle debug plan > out.first-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+trace print_requests.py //serving-endpoints | jq --sort-keys > out.first-requests.$DATABRICKS_BUNDLE_ENGINE.json
+
+echo "$ENDPOINT_NAME:ENDPOINT_NAME_1" >> ACC_REPLS
+
+# Record the endpoint ID for the first endpoint
+get_output=$($CLI serving-endpoints get $ENDPOINT_NAME)
+endpoint_id=$(echo "$get_output" | jq -r '.id')
+echo "$endpoint_id:ENDPOINT_ID_1" >> ACC_REPLS
+
+title "Print the GET endpoint details\n"
+echo "$get_output" | jq '{name, creator}'
+
+export ENDPOINT_NAME="test-endpoint-$UNIQUE_NAME-2"
+envsubst < databricks.yml.tmpl > databricks.yml
+trace $CLI bundle debug plan > out.second-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+trace print_requests.py //serving-endpoints | jq --sort-keys > out.second-requests.$DATABRICKS_BUNDLE_ENGINE.json
+
+echo "$ENDPOINT_NAME:ENDPOINT_NAME_2" >> ACC_REPLS
+
+# Record the endpoint ID for the second endpoint
+get_output=$($CLI serving-endpoints get $ENDPOINT_NAME)
+endpoint_id=$(echo "$get_output" | jq -r '.id')
+echo "$endpoint_id:ENDPOINT_ID_2" >> ACC_REPLS
+
+title "Print the GET endpoint details\n"
+echo "$get_output" | jq '{name, creator}'
diff --git a/acceptance/bundle/resources/model_serving_endpoints/basic/test.toml b/acceptance/bundle/resources/model_serving_endpoints/basic/test.toml
new file mode 100644
index 0000000000..c64c7dfec0
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/basic/test.toml
@@ -0,0 +1,6 @@
+Local = false
+Cloud = true
+RecordRequests = true
+
+[EnvMatrix]
+  DATABRICKS_BUNDLE_ENGINE = ["terraform", "direct"]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/databricks.yml.tmpl b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/databricks.yml.tmpl
new file mode 100644
index 0000000000..5cb6657fd5
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/databricks.yml.tmpl
@@ -0,0 +1,23 @@
+bundle:
+  name: test-mse-catalog-$UNIQUE_NAME
+
+workspace:
+  root_path: ~/.bundle/$UNIQUE_NAME
+
+resources:
+  model_serving_endpoints:
+    test_endpoint:
+      name: test-endpoint-$UNIQUE_NAME
+      config:
+        served_entities:
+          - name: prod
+            external_model:
+              name: gpt-4o-mini
+              provider: openai
+              task: llm/v1/chat
+              openai_config:
+                openai_api_key: "{{secrets/test-scope/openai-key}}"
+        auto_capture_config:
+          catalog_name: main
+          schema_name: default
+          table_name_prefix: my_table
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.first-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.first-plan.direct.json
new file mode 100644
index 0000000000..11afeef35c
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.first-plan.direct.json
@@ -0,0 +1,32 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create",
+      "new_state": {
+        "config": {
+          "config": {
+            "auto_capture_config": {
+              "catalog_name": "main",
+              "schema_name": "default",
+              "table_name_prefix": "my_table"
+            },
+            "served_entities": [
+              {
+                "external_model": {
+                  "name": "gpt-4o-mini",
+                  "openai_config": {
+                    "openai_api_key": "{{secrets/test-scope/openai-key}}"
+                  },
+                  "provider": "openai",
+                  "task": "llm/v1/chat"
+                },
+                "name": "prod"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]"
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.first-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.first-plan.terraform.json
new file mode 100644
index 0000000000..82ea9497c9
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.first-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.second-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.second-plan.direct.json
new file mode 100644
index 0000000000..156261de42
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.second-plan.direct.json
@@ -0,0 +1,39 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate",
+      "new_state": {
+        "config": {
+          "config": {
+            "auto_capture_config": {
+              "catalog_name": "other_catalog",
+              "schema_name": "default",
+              "table_name_prefix": "my_table"
+            },
+            "served_entities": [
+              {
+                "external_model": {
+                  "name": "gpt-4o-mini",
+                  "openai_config": {
+                    "openai_api_key": "{{secrets/test-scope/openai-key}}"
+                  },
+                  "provider": "openai",
+                  "task": "llm/v1/chat"
+                },
+                "name": "prod"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]"
+        }
+      },
+      "changes": {
+        "local": {
+          "config.auto_capture_config.catalog_name": {
+            "action": "recreate"
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.second-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.second-plan.terraform.json
new file mode 100644
index 0000000000..225f2eb49a
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.second-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.test.toml b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.test.toml
new file mode 100644
index 0000000000..d560f1de04
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/out.test.toml
@@ -0,0 +1,5 @@
+Local = true
+Cloud = false
+
+[EnvMatrix]
+  DATABRICKS_BUNDLE_ENGINE = ["terraform", "direct"]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/output.txt b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/output.txt
new file mode 100644
index 0000000000..8688f1beee
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/output.txt
@@ -0,0 +1,95 @@
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "auto_capture_config": {
+        "catalog_name": "main",
+        "schema_name": "default",
+        "table_name_prefix": "my_table"
+      },
+      "served_entities": [
+        {
+          "external_model": {
+            "name": "gpt-4o-mini",
+            "openai_config": {
+              "openai_api_key": "{{secrets/test-scope/openai-key}}"
+            },
+            "provider": "openai",
+            "task": "llm/v1/chat"
+          },
+          "name": "prod"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]"
+  }
+}
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+"main"
+
+>>> update_file.py databricks.yml catalog_name: main catalog_name: other_catalog
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "DELETE",
+  "path": "/api/2.0/serving-endpoints/[ORIGINAL_ENDPOINT_ID]"
+}
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "auto_capture_config": {
+        "catalog_name": "other_catalog",
+        "schema_name": "default",
+        "table_name_prefix": "my_table"
+      },
+      "served_entities": [
+        {
+          "external_model": {
+            "name": "gpt-4o-mini",
+            "openai_config": {
+              "openai_api_key": "{{secrets/test-scope/openai-key}}"
+            },
+            "provider": "openai",
+            "task": "llm/v1/chat"
+          },
+          "name": "prod"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]"
+  }
+}
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+"other_catalog"
+
+>>> [CLI] bundle destroy --auto-approve
+The following resources will be deleted:
+  delete model_serving_endpoint test_endpoint
+
+All files and directories at the following location will be deleted: /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]
+
+Deleting files...
+Destroy complete!
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/script b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/script
new file mode 100755
index 0000000000..e4a7b4276a
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/catalog-name/script
@@ -0,0 +1,26 @@
+envsubst < databricks.yml.tmpl > databricks.yml
+
+cleanup() {
+    trace $CLI bundle destroy --auto-approve
+    rm -f out.requests.txt
+}
+trap cleanup EXIT
+
+trace $CLI bundle debug plan > out.first-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$ENDPOINT_ID:ORIGINAL_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${ENDPOINT_ID}" | jq '.config.auto_capture_config.catalog_name'
+
+trace update_file.py databricks.yml "catalog_name: main" "catalog_name: other_catalog"
+trace $CLI bundle debug plan > out.second-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+NEW_ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$NEW_ENDPOINT_ID:NEW_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${NEW_ENDPOINT_ID}" | jq '.config.auto_capture_config.catalog_name'
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/databricks.yml.tmpl b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/databricks.yml.tmpl
new file mode 100644
index 0000000000..fe2f5f02c9
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/databricks.yml.tmpl
@@ -0,0 +1,19 @@
+bundle:
+  name: test-mse-name-$UNIQUE_NAME
+
+workspace:
+  root_path: ~/.bundle/$UNIQUE_NAME
+
+resources:
+  model_serving_endpoints:
+    test_endpoint:
+      name: test-endpoint-$UNIQUE_NAME
+      config:
+        served_entities:
+          - name: prod
+            external_model:
+              name: gpt-4o-mini
+              provider: openai
+              task: llm/v1/chat
+              openai_config:
+                openai_api_key: "{{secrets/test-scope/openai-key}}"
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.first-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.first-plan.direct.json
new file mode 100644
index 0000000000..ece5a9db9a
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.first-plan.direct.json
@@ -0,0 +1,27 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create",
+      "new_state": {
+        "config": {
+          "config": {
+            "served_entities": [
+              {
+                "external_model": {
+                  "name": "gpt-4o-mini",
+                  "openai_config": {
+                    "openai_api_key": "{{secrets/test-scope/openai-key}}"
+                  },
+                  "provider": "openai",
+                  "task": "llm/v1/chat"
+                },
+                "name": "prod"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]"
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.first-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.first-plan.terraform.json
new file mode 100644
index 0000000000..82ea9497c9
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.first-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.second-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.second-plan.direct.json
new file mode 100644
index 0000000000..fa0d636f23
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.second-plan.direct.json
@@ -0,0 +1,34 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate",
+      "new_state": {
+        "config": {
+          "config": {
+            "served_entities": [
+              {
+                "external_model": {
+                  "name": "gpt-4o-mini",
+                  "openai_config": {
+                    "openai_api_key": "{{secrets/test-scope/openai-key}}"
+                  },
+                  "provider": "openai",
+                  "task": "llm/v1/chat"
+                },
+                "name": "prod"
+              }
+            ]
+          },
+          "name": "[NEW_ENDPOINT_ID]"
+        }
+      },
+      "changes": {
+        "local": {
+          "name": {
+            "action": "recreate"
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.second-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.second-plan.terraform.json
new file mode 100644
index 0000000000..225f2eb49a
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.second-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.test.toml b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.test.toml
new file mode 100644
index 0000000000..d560f1de04
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/out.test.toml
@@ -0,0 +1,5 @@
+Local = true
+Cloud = false
+
+[EnvMatrix]
+  DATABRICKS_BUNDLE_ENGINE = ["terraform", "direct"]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/output.txt b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/output.txt
new file mode 100644
index 0000000000..70320f337f
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/output.txt
@@ -0,0 +1,85 @@
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "served_entities": [
+        {
+          "external_model": {
+            "name": "gpt-4o-mini",
+            "openai_config": {
+              "openai_api_key": "{{secrets/test-scope/openai-key}}"
+            },
+            "provider": "openai",
+            "task": "llm/v1/chat"
+          },
+          "name": "prod"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]"
+  }
+}
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+"[ORIGINAL_ENDPOINT_ID]"
+
+>>> update_file.py databricks.yml name: [ORIGINAL_ENDPOINT_ID] name: [NEW_ENDPOINT_ID]
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "DELETE",
+  "path": "/api/2.0/serving-endpoints/[ORIGINAL_ENDPOINT_ID]"
+}
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "served_entities": [
+        {
+          "external_model": {
+            "name": "gpt-4o-mini",
+            "openai_config": {
+              "openai_api_key": "{{secrets/test-scope/openai-key}}"
+            },
+            "provider": "openai",
+            "task": "llm/v1/chat"
+          },
+          "name": "prod"
+        }
+      ]
+    },
+    "name": "[NEW_ENDPOINT_ID]"
+  }
+}
+
+>>> [CLI] serving-endpoints get [NEW_ENDPOINT_ID]
+"[NEW_ENDPOINT_ID]"
+
+>>> [CLI] bundle destroy --auto-approve
+The following resources will be deleted:
+  delete model_serving_endpoint test_endpoint
+
+All files and directories at the following location will be deleted: /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]
+
+Deleting files...
+Destroy complete!
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/script b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/script
new file mode 100755
index 0000000000..fdbbeaa40d
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/name-change/script
@@ -0,0 +1,26 @@
+envsubst < databricks.yml.tmpl > databricks.yml
+
+cleanup() {
+    trace $CLI bundle destroy --auto-approve
+    rm -f out.requests.txt
+}
+trap cleanup EXIT
+
+trace $CLI bundle debug plan > out.first-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$ENDPOINT_ID:ORIGINAL_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${ENDPOINT_ID}" | jq '.name'
+
+trace update_file.py databricks.yml "name: test-endpoint-$UNIQUE_NAME" "name: test-endpoint-2-$UNIQUE_NAME"
+trace $CLI bundle debug plan > out.second-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+NEW_ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$NEW_ENDPOINT_ID:NEW_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${NEW_ENDPOINT_ID}" | jq '.name'
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/databricks.yml.tmpl b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/databricks.yml.tmpl
new file mode 100644
index 0000000000..d20ffc62da
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/databricks.yml.tmpl
@@ -0,0 +1,18 @@
+bundle:
+  name: test-mse-route-$UNIQUE_NAME
+
+workspace:
+  root_path: ~/.bundle/$UNIQUE_NAME
+
+resources:
+  model_serving_endpoints:
+    test_endpoint:
+      name: test-endpoint-$UNIQUE_NAME
+      config:
+        served_entities:
+          - name: "llama"
+            entity_name: system.ai.llama_v3_2_1b_instruct
+            entity_version: 1
+            workload_size: Small
+            scale_to_zero_enabled: true
+      route_optimized: false
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-get.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-get.direct.json
new file mode 100644
index 0000000000..f74ee87921
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-get.direct.json
@@ -0,0 +1,4 @@
+{
+  "route_optimized": false,
+  "id": "[UUID]"
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-get.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-get.terraform.json
new file mode 100644
index 0000000000..f4d7f8991f
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-get.terraform.json
@@ -0,0 +1,4 @@
+{
+  "route_optimized": null,
+  "id": "[UUID]"
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-plan.direct.json
new file mode 100644
index 0000000000..c11b2e1980
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-plan.direct.json
@@ -0,0 +1,24 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create",
+      "new_state": {
+        "config": {
+          "config": {
+            "served_entities": [
+              {
+                "entity_name": "system.ai.llama_v3_2_1b_instruct",
+                "entity_version": "1",
+                "name": "llama",
+                "scale_to_zero_enabled": true,
+                "workload_size": "Small"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]",
+          "route_optimized": false
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-plan.terraform.json
new file mode 100644
index 0000000000..82ea9497c9
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-requests.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-requests.direct.json
new file mode 100644
index 0000000000..a287e1d81e
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-requests.direct.json
@@ -0,0 +1,19 @@
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "served_entities": [
+        {
+          "entity_name": "system.ai.llama_v3_2_1b_instruct",
+          "entity_version": "1",
+          "name": "llama",
+          "scale_to_zero_enabled": true,
+          "workload_size": "Small"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]",
+    "route_optimized": false
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-requests.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-requests.terraform.json
new file mode 100644
index 0000000000..95e359ba4e
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.first-requests.terraform.json
@@ -0,0 +1,18 @@
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "served_entities": [
+        {
+          "entity_name": "system.ai.llama_v3_2_1b_instruct",
+          "entity_version": "1",
+          "name": "llama",
+          "scale_to_zero_enabled": true,
+          "workload_size": "Small"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]"
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.second-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.second-plan.direct.json
new file mode 100644
index 0000000000..145e2879d7
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.second-plan.direct.json
@@ -0,0 +1,31 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate",
+      "new_state": {
+        "config": {
+          "config": {
+            "served_entities": [
+              {
+                "entity_name": "system.ai.llama_v3_2_1b_instruct",
+                "entity_version": "1",
+                "name": "llama",
+                "scale_to_zero_enabled": true,
+                "workload_size": "Small"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]",
+          "route_optimized": true
+        }
+      },
+      "changes": {
+        "local": {
+          "route_optimized": {
+            "action": "recreate"
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.second-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.second-plan.terraform.json
new file mode 100644
index 0000000000..225f2eb49a
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.second-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.test.toml b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.test.toml
new file mode 100644
index 0000000000..d560f1de04
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/out.test.toml
@@ -0,0 +1,5 @@
+Local = true
+Cloud = false
+
+[EnvMatrix]
+  DATABRICKS_BUNDLE_ENGINE = ["terraform", "direct"]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/output.txt b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/output.txt
new file mode 100644
index 0000000000..8c29071412
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/output.txt
@@ -0,0 +1,76 @@
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+
+>>> update_file.py databricks.yml route_optimized: false route_optimized: true
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "DELETE",
+  "path": "/api/2.0/serving-endpoints/[ORIGINAL_ENDPOINT_ID]"
+}
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "served_entities": [
+        {
+          "entity_name": "system.ai.llama_v3_2_1b_instruct",
+          "entity_version": "1",
+          "name": "llama",
+          "scale_to_zero_enabled": true,
+          "workload_size": "Small"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]",
+    "route_optimized": true
+  }
+}
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+{
+  "config": {
+    "served_entities": [
+      {
+        "entity_name":"system.ai.llama_v3_2_1b_instruct",
+        "entity_version":"1",
+        "name":"llama",
+        "scale_to_zero_enabled":true,
+        "workload_size":"Small"
+      }
+    ]
+  },
+  "creator":"[USERNAME]",
+  "id":"[UUID]",
+  "name":"[ORIGINAL_ENDPOINT_ID]",
+  "route_optimized":true,
+  "state": {
+    "config_update":"NOT_UPDATING"
+  }
+}
+
+>>> [CLI] bundle destroy --auto-approve
+The following resources will be deleted:
+  delete model_serving_endpoint test_endpoint
+
+All files and directories at the following location will be deleted: /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]
+
+Deleting files...
+Destroy complete!
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/script b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/script
new file mode 100755
index 0000000000..fb93230a9e
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/route-optimized/script
@@ -0,0 +1,26 @@
+envsubst < databricks.yml.tmpl > databricks.yml
+
+cleanup() {
+    trace $CLI bundle destroy --auto-approve
+    rm -f out.requests.txt
+}
+trap cleanup EXIT
+
+trace $CLI bundle debug plan > out.first-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$ENDPOINT_ID:ORIGINAL_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${ENDPOINT_ID}" | jq '{route_optimized, id}' > out.first-get.$DATABRICKS_BUNDLE_ENGINE.json
+
+print_requests.py //serving-endpoints > out.first-requests.$DATABRICKS_BUNDLE_ENGINE.json
+
+trace update_file.py databricks.yml "route_optimized: false" "route_optimized: true"
+trace $CLI bundle debug plan > out.second-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+# New and original endpoint ID should remain the same since they are both the name of the endpoint.
+NEW_ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+trace $CLI serving-endpoints get "${NEW_ENDPOINT_ID}"
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/databricks.yml.tmpl b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/databricks.yml.tmpl
new file mode 100644
index 0000000000..f57fcc4f03
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/databricks.yml.tmpl
@@ -0,0 +1,23 @@
+bundle:
+  name: test-mse-schema-$UNIQUE_NAME
+
+workspace:
+  root_path: ~/.bundle/$UNIQUE_NAME
+
+resources:
+  model_serving_endpoints:
+    test_endpoint:
+      name: test-endpoint-$UNIQUE_NAME
+      config:
+        served_entities:
+          - name: prod
+            external_model:
+              name: gpt-4o-mini
+              provider: openai
+              task: llm/v1/chat
+              openai_config:
+                openai_api_key: "{{secrets/test-scope/openai-key}}"
+        auto_capture_config:
+          catalog_name: main
+          schema_name: default
+          table_name_prefix: my_table
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.first-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.first-plan.direct.json
new file mode 100644
index 0000000000..11afeef35c
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.first-plan.direct.json
@@ -0,0 +1,32 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create",
+      "new_state": {
+        "config": {
+          "config": {
+            "auto_capture_config": {
+              "catalog_name": "main",
+              "schema_name": "default",
+              "table_name_prefix": "my_table"
+            },
+            "served_entities": [
+              {
+                "external_model": {
+                  "name": "gpt-4o-mini",
+                  "openai_config": {
+                    "openai_api_key": "{{secrets/test-scope/openai-key}}"
+                  },
+                  "provider": "openai",
+                  "task": "llm/v1/chat"
+                },
+                "name": "prod"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]"
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.first-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.first-plan.terraform.json
new file mode 100644
index 0000000000..82ea9497c9
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.first-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.second-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.second-plan.direct.json
new file mode 100644
index 0000000000..c1f8fa4b13
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.second-plan.direct.json
@@ -0,0 +1,39 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate",
+      "new_state": {
+        "config": {
+          "config": {
+            "auto_capture_config": {
+              "catalog_name": "main",
+              "schema_name": "other_schema",
+              "table_name_prefix": "my_table"
+            },
+            "served_entities": [
+              {
+                "external_model": {
+                  "name": "gpt-4o-mini",
+                  "openai_config": {
+                    "openai_api_key": "{{secrets/test-scope/openai-key}}"
+                  },
+                  "provider": "openai",
+                  "task": "llm/v1/chat"
+                },
+                "name": "prod"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]"
+        }
+      },
+      "changes": {
+        "local": {
+          "config.auto_capture_config.schema_name": {
+            "action": "recreate"
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.second-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.second-plan.terraform.json
new file mode 100644
index 0000000000..225f2eb49a
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.second-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.test.toml b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.test.toml
new file mode 100644
index 0000000000..d560f1de04
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/out.test.toml
@@ -0,0 +1,5 @@
+Local = true
+Cloud = false
+
+[EnvMatrix]
+  DATABRICKS_BUNDLE_ENGINE = ["terraform", "direct"]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/output.txt b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/output.txt
new file mode 100644
index 0000000000..edf577a5c4
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/output.txt
@@ -0,0 +1,95 @@
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "auto_capture_config": {
+        "catalog_name": "main",
+        "schema_name": "default",
+        "table_name_prefix": "my_table"
+      },
+      "served_entities": [
+        {
+          "external_model": {
+            "name": "gpt-4o-mini",
+            "openai_config": {
+              "openai_api_key": "{{secrets/test-scope/openai-key}}"
+            },
+            "provider": "openai",
+            "task": "llm/v1/chat"
+          },
+          "name": "prod"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]"
+  }
+}
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+"default"
+
+>>> update_file.py databricks.yml schema_name: default schema_name: other_schema
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "DELETE",
+  "path": "/api/2.0/serving-endpoints/[ORIGINAL_ENDPOINT_ID]"
+}
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "auto_capture_config": {
+        "catalog_name": "main",
+        "schema_name": "other_schema",
+        "table_name_prefix": "my_table"
+      },
+      "served_entities": [
+        {
+          "external_model": {
+            "name": "gpt-4o-mini",
+            "openai_config": {
+              "openai_api_key": "{{secrets/test-scope/openai-key}}"
+            },
+            "provider": "openai",
+            "task": "llm/v1/chat"
+          },
+          "name": "prod"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]"
+  }
+}
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+"other_schema"
+
+>>> [CLI] bundle destroy --auto-approve
+The following resources will be deleted:
+  delete model_serving_endpoint test_endpoint
+
+All files and directories at the following location will be deleted: /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]
+
+Deleting files...
+Destroy complete!
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/script b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/script
new file mode 100755
index 0000000000..59c22a077a
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/schema-name/script
@@ -0,0 +1,26 @@
+envsubst < databricks.yml.tmpl > databricks.yml
+
+cleanup() {
+    trace $CLI bundle destroy --auto-approve
+    rm -f out.requests.txt
+}
+trap cleanup EXIT
+
+trace $CLI bundle debug plan > out.first-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$ENDPOINT_ID:ORIGINAL_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${ENDPOINT_ID}" | jq '.config.auto_capture_config.schema_name'
+
+trace update_file.py databricks.yml "schema_name: default" "schema_name: other_schema"
+trace $CLI bundle debug plan > out.second-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+NEW_ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$NEW_ENDPOINT_ID:NEW_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${NEW_ENDPOINT_ID}" | jq '.config.auto_capture_config.schema_name'
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/databricks.yml.tmpl b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/databricks.yml.tmpl
new file mode 100644
index 0000000000..8b22b900f6
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/databricks.yml.tmpl
@@ -0,0 +1,23 @@
+bundle:
+  name: test-mse-table-$UNIQUE_NAME
+
+workspace:
+  root_path: ~/.bundle/$UNIQUE_NAME
+
+resources:
+  model_serving_endpoints:
+    test_endpoint:
+      name: test-endpoint-$UNIQUE_NAME
+      config:
+        served_entities:
+          - name: prod
+            external_model:
+              name: gpt-4o-mini
+              provider: openai
+              task: llm/v1/chat
+              openai_config:
+                openai_api_key: "{{secrets/test-scope/openai-key}}"
+        auto_capture_config:
+          catalog_name: main
+          schema_name: default
+          table_name_prefix: my_table
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.first-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.first-plan.direct.json
new file mode 100644
index 0000000000..11afeef35c
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.first-plan.direct.json
@@ -0,0 +1,32 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create",
+      "new_state": {
+        "config": {
+          "config": {
+            "auto_capture_config": {
+              "catalog_name": "main",
+              "schema_name": "default",
+              "table_name_prefix": "my_table"
+            },
+            "served_entities": [
+              {
+                "external_model": {
+                  "name": "gpt-4o-mini",
+                  "openai_config": {
+                    "openai_api_key": "{{secrets/test-scope/openai-key}}"
+                  },
+                  "provider": "openai",
+                  "task": "llm/v1/chat"
+                },
+                "name": "prod"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]"
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.first-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.first-plan.terraform.json
new file mode 100644
index 0000000000..82ea9497c9
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.first-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "create"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.second-plan.direct.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.second-plan.direct.json
new file mode 100644
index 0000000000..c546a1392b
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.second-plan.direct.json
@@ -0,0 +1,39 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate",
+      "new_state": {
+        "config": {
+          "config": {
+            "auto_capture_config": {
+              "catalog_name": "main",
+              "schema_name": "default",
+              "table_name_prefix": "other_table"
+            },
+            "served_entities": [
+              {
+                "external_model": {
+                  "name": "gpt-4o-mini",
+                  "openai_config": {
+                    "openai_api_key": "{{secrets/test-scope/openai-key}}"
+                  },
+                  "provider": "openai",
+                  "task": "llm/v1/chat"
+                },
+                "name": "prod"
+              }
+            ]
+          },
+          "name": "[ORIGINAL_ENDPOINT_ID]"
+        }
+      },
+      "changes": {
+        "local": {
+          "config.auto_capture_config.table_name_prefix": {
+            "action": "recreate"
+          }
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.second-plan.terraform.json b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.second-plan.terraform.json
new file mode 100644
index 0000000000..225f2eb49a
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.second-plan.terraform.json
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.test_endpoint": {
+      "action": "recreate"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.test.toml b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.test.toml
new file mode 100644
index 0000000000..d560f1de04
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/out.test.toml
@@ -0,0 +1,5 @@
+Local = true
+Cloud = false
+
+[EnvMatrix]
+  DATABRICKS_BUNDLE_ENGINE = ["terraform", "direct"]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/output.txt b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/output.txt
new file mode 100644
index 0000000000..b178c70fb3
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/output.txt
@@ -0,0 +1,95 @@
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "auto_capture_config": {
+        "catalog_name": "main",
+        "schema_name": "default",
+        "table_name_prefix": "my_table"
+      },
+      "served_entities": [
+        {
+          "external_model": {
+            "name": "gpt-4o-mini",
+            "openai_config": {
+              "openai_api_key": "{{secrets/test-scope/openai-key}}"
+            },
+            "provider": "openai",
+            "task": "llm/v1/chat"
+          },
+          "name": "prod"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]"
+  }
+}
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+"my_table"
+
+>>> update_file.py databricks.yml table_name_prefix: my_table table_name_prefix: other_table
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "DELETE",
+  "path": "/api/2.0/serving-endpoints/[ORIGINAL_ENDPOINT_ID]"
+}
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "auto_capture_config": {
+        "catalog_name": "main",
+        "schema_name": "default",
+        "table_name_prefix": "other_table"
+      },
+      "served_entities": [
+        {
+          "external_model": {
+            "name": "gpt-4o-mini",
+            "openai_config": {
+              "openai_api_key": "{{secrets/test-scope/openai-key}}"
+            },
+            "provider": "openai",
+            "task": "llm/v1/chat"
+          },
+          "name": "prod"
+        }
+      ]
+    },
+    "name": "[ORIGINAL_ENDPOINT_ID]"
+  }
+}
+
+>>> [CLI] serving-endpoints get [ORIGINAL_ENDPOINT_ID]
+"other_table"
+
+>>> [CLI] bundle destroy --auto-approve
+The following resources will be deleted:
+  delete model_serving_endpoint test_endpoint
+
+All files and directories at the following location will be deleted: /Workspace/Users/[USERNAME]/.bundle/[UNIQUE_NAME]
+
+Deleting files...
+Destroy complete!
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/script b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/script
new file mode 100755
index 0000000000..1d3f7b35a0
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/table-prefix/script
@@ -0,0 +1,26 @@
+envsubst < databricks.yml.tmpl > databricks.yml
+
+cleanup() {
+    trace $CLI bundle destroy --auto-approve
+    rm -f out.requests.txt
+}
+trap cleanup EXIT
+
+trace $CLI bundle debug plan > out.first-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$ENDPOINT_ID:ORIGINAL_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${ENDPOINT_ID}" | jq '.config.auto_capture_config.table_name_prefix'
+
+trace update_file.py databricks.yml "table_name_prefix: my_table" "table_name_prefix: other_table"
+trace $CLI bundle debug plan > out.second-plan.$DATABRICKS_BUNDLE_ENGINE.json
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+NEW_ENDPOINT_ID=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.test_endpoint.id')
+echo "$NEW_ENDPOINT_ID:NEW_ENDPOINT_ID" >> ACC_REPLS
+trace $CLI serving-endpoints get "${NEW_ENDPOINT_ID}" | jq '.config.auto_capture_config.table_name_prefix'
diff --git a/acceptance/bundle/resources/model_serving_endpoints/recreate/test.toml b/acceptance/bundle/resources/model_serving_endpoints/recreate/test.toml
new file mode 100644
index 0000000000..3aa4a19050
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/recreate/test.toml
@@ -0,0 +1,7 @@
+Local = true
+Cloud = false
+RecordRequests = true
+
+Ignore = [
+    "databricks.yml",
+]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/databricks.yml.tmpl b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/databricks.yml.tmpl
new file mode 100644
index 0000000000..73e0943152
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/databricks.yml.tmpl
@@ -0,0 +1,14 @@
+bundle:
+  name: acc-$UNIQUE_NAME
+
+resources:
+  model_serving_endpoints:
+    my_endpoint:
+      name: test-endpoint-$UNIQUE_NAME
+      config:
+        served_entities:
+          - name: "llama"
+            entity_name: system.ai.llama_v3_2_1b_instruct
+            entity_version: $VERSION
+            workload_size: Small
+            scale_to_zero_enabled: true
diff --git a/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.plan.direct.txt b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.plan.direct.txt
new file mode 100644
index 0000000000..3539d2da56
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.plan.direct.txt
@@ -0,0 +1,23 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.my_endpoint": {
+      "action": "create",
+      "new_state": {
+        "config": {
+          "config": {
+            "served_entities": [
+              {
+                "entity_name": "system.ai.llama_v3_2_1b_instruct",
+                "entity_version": "[VERSION]",
+                "name": "llama",
+                "scale_to_zero_enabled": true,
+                "workload_size": "Small"
+              }
+            ]
+          },
+          "name": "test-endpoint-[UNIQUE_NAME]"
+        }
+      }
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.plan.terraform.txt b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.plan.terraform.txt
new file mode 100644
index 0000000000..1d92fcbea4
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.plan.terraform.txt
@@ -0,0 +1,7 @@
+{
+  "plan": {
+    "resources.model_serving_endpoints.my_endpoint": {
+      "action": "create"
+    }
+  }
+}
diff --git a/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.test.toml b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.test.toml
new file mode 100644
index 0000000000..5366fbb1a4
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/out.test.toml
@@ -0,0 +1,6 @@
+Local = true
+Cloud = true
+CloudSlow = true
+
+[EnvMatrix]
+  DATABRICKS_BUNDLE_ENGINE = ["terraform", "direct"]
diff --git a/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/output.txt b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/output.txt
new file mode 100644
index 0000000000..1fd0ab1517
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/output.txt
@@ -0,0 +1,43 @@
+
+>>> [CLI] bundle debug plan
+
+>>> [CLI] bundle deploy
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/acc-[UNIQUE_NAME]/default/files...
+Deploying resources...
+Updating deployment state...
+Deployment complete!
+
+>>> print_requests.py //serving-endpoints
+{
+  "method": "POST",
+  "path": "/api/2.0/serving-endpoints",
+  "body": {
+    "config": {
+      "served_entities": [
+        {
+          "entity_name": "system.ai.llama_v3_2_1b_instruct",
+          "entity_version": "[VERSION]",
+          "name": "llama",
+          "scale_to_zero_enabled": true,
+          "workload_size": "Small"
+        }
+      ]
+    },
+    "name": "test-endpoint-[UNIQUE_NAME]"
+  }
+}
+
+>>> [CLI] serving-endpoints get test-endpoint-[UNIQUE_NAME]
+{
+  "name": "test-endpoint-[UNIQUE_NAME]",
+  "creator": "[USERNAME]"
+}
+
+>>> [CLI] bundle destroy --auto-approve
+The following resources will be deleted:
+  delete model_serving_endpoint my_endpoint
+
+All files and directories at the following location will be deleted: /Workspace/Users/[USERNAME]/.bundle/acc-[UNIQUE_NAME]/default
+
+Deleting files...
+Destroy complete!
diff --git a/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/script b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/script
new file mode 100644
index 0000000000..fab464d557
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/script
@@ -0,0 +1,17 @@
+export VERSION=$($CLI model-versions list system.ai.llama_v3_2_1b_instruct | jq '.[0].version')
+
+envsubst < databricks.yml.tmpl > databricks.yml
+
+cleanup() {
+    trace $CLI bundle destroy --auto-approve
+    rm out.requests.txt
+}
+trap cleanup EXIT
+
+trace $CLI bundle debug plan > out.plan.$DATABRICKS_BUNDLE_ENGINE.txt
+trace $CLI bundle deploy
+
+trace print_requests.py //serving-endpoints
+
+endpoint_name=$($CLI bundle summary -o json | jq -r '.resources.model_serving_endpoints.my_endpoint.name')
+trace $CLI serving-endpoints get $endpoint_name | jq '{name, creator}'
diff --git a/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/test.toml b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/test.toml
new file mode 100644
index 0000000000..b3c25f76e7
--- /dev/null
+++ b/acceptance/bundle/resources/model_serving_endpoints/running-endpoint/test.toml
@@ -0,0 +1,20 @@
+Timeout = '30m'
+Cloud = true
+Local = true
+CloudSlow = true
+
+[[Server]]
+Pattern = "GET /api/2.1/unity-catalog/models/system.ai.llama_v3_2_1b_instruct/versions"
+Response.Body = '''
+{
+  "model_versions": [
+    {
+      "version": 1
+    }
+  ]
+}
+'''
+
+[[Repls]]
+Old = '"entity_version": "1",'
+New = '"entity_version": "[VERSION]",'
diff --git a/bundle/direct/dresources/all.go b/bundle/direct/dresources/all.go
index 5a66fe9132..5fb8e95a58 100644
--- a/bundle/direct/dresources/all.go
+++ b/bundle/direct/dresources/all.go
@@ -7,31 +7,33 @@ import (
 )
 
 var SupportedResources = map[string]any{
-	"jobs":                   (*ResourceJob)(nil),
-	"pipelines":              (*ResourcePipeline)(nil),
-	"experiments":            (*ResourceExperiment)(nil),
-	"schemas":                (*ResourceSchema)(nil),
-	"volumes":                (*ResourceVolume)(nil),
-	"models":                 (*ResourceMlflowModel)(nil),
-	"apps":                   (*ResourceApp)(nil),
-	"sql_warehouses":         (*ResourceSqlWarehouse)(nil),
-	"database_instances":     (*ResourceDatabaseInstance)(nil),
-	"database_catalogs":      (*ResourceDatabaseCatalog)(nil),
-	"synced_database_tables": (*ResourceSyncedDatabaseTable)(nil),
-	"alerts":                 (*ResourceAlert)(nil),
-	"clusters":               (*ResourceCluster)(nil),
-	"registered_models":      (*ResourceRegisteredModel)(nil),
-	"dashboards":             (*ResourceDashboard)(nil),
+	"jobs":                    (*ResourceJob)(nil),
+	"pipelines":               (*ResourcePipeline)(nil),
+	"experiments":             (*ResourceExperiment)(nil),
+	"schemas":                 (*ResourceSchema)(nil),
+	"volumes":                 (*ResourceVolume)(nil),
+	"models":                  (*ResourceMlflowModel)(nil),
+	"apps":                    (*ResourceApp)(nil),
+	"sql_warehouses":          (*ResourceSqlWarehouse)(nil),
+	"database_instances":      (*ResourceDatabaseInstance)(nil),
+	"database_catalogs":       (*ResourceDatabaseCatalog)(nil),
+	"synced_database_tables":  (*ResourceSyncedDatabaseTable)(nil),
+	"alerts":                  (*ResourceAlert)(nil),
+	"clusters":                (*ResourceCluster)(nil),
+	"registered_models":       (*ResourceRegisteredModel)(nil),
+	"dashboards":              (*ResourceDashboard)(nil),
+	"model_serving_endpoints": (*ResourceModelServingEndpoint)(nil),
 
 	// Permissions
-	"jobs.permissions":               (*ResourcePermissions)(nil),
-	"pipelines.permissions":          (*ResourcePermissions)(nil),
-	"apps.permissions":               (*ResourcePermissions)(nil),
-	"clusters.permissions":           (*ResourcePermissions)(nil),
-	"database_instances.permissions": (*ResourcePermissions)(nil),
-	"experiments.permissions":        (*ResourcePermissions)(nil),
-	"models.permissions":             (*ResourcePermissions)(nil),
-	"sql_warehouses.permissions":     (*ResourcePermissions)(nil),
+	"jobs.permissions":                    (*ResourcePermissions)(nil),
+	"pipelines.permissions":               (*ResourcePermissions)(nil),
+	"apps.permissions":                    (*ResourcePermissions)(nil),
+	"clusters.permissions":                (*ResourcePermissions)(nil),
+	"database_instances.permissions":      (*ResourcePermissions)(nil),
+	"experiments.permissions":             (*ResourcePermissions)(nil),
+	"models.permissions":                  (*ResourcePermissions)(nil),
+	"sql_warehouses.permissions":          (*ResourcePermissions)(nil),
+	"model_serving_endpoints.permissions": (*ResourcePermissions)(nil),
 
 	// Grants
 	"schemas.grants":           (*ResourceGrants)(nil),
diff --git a/bundle/direct/dresources/all_test.go b/bundle/direct/dresources/all_test.go
index 08ac7b7e51..506be36c47 100644
--- a/bundle/direct/dresources/all_test.go
+++ b/bundle/direct/dresources/all_test.go
@@ -24,6 +24,7 @@ import (
 	"github.com/databricks/databricks-sdk-go/service/jobs"
 	"github.com/databricks/databricks-sdk-go/service/ml"
 	"github.com/databricks/databricks-sdk-go/service/pipelines"
+	"github.com/databricks/databricks-sdk-go/service/serving"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
@@ -97,6 +98,50 @@ var testConfig map[string]any = map[string]any{
 			},
 		},
 	},
+
+	"model_serving_endpoints": &resources.ModelServingEndpoint{
+		CreateServingEndpoint: serving.CreateServingEndpoint{
+			Name: "my-endpoint",
+			Config: &serving.EndpointCoreConfigInput{
+				Name: "my-endpoint",
+				AutoCaptureConfig: &serving.AutoCaptureConfigInput{
+					CatalogName:     "main",
+					SchemaName:      "myschema",
+					TableNamePrefix: "my_table",
+					Enabled:         true,
+					ForceSendFields: nil,
+				},
+				ServedModels: nil,
+				ServedEntities: []serving.ServedEntityInput{
+					{
+						EntityName:                "entity-name",
+						EntityVersion:             "1",
+						WorkloadSize:              "Small",
+						ScaleToZeroEnabled:        true,
+						WorkloadType:              serving.ServingModelWorkloadTypeCpu,
+						EnvironmentVars:           map[string]string{"key": "value"},
+						InstanceProfileArn:        "arn:aws:iam::123456789012:instance-profile/my-instance-profile",
+						MaxProvisionedConcurrency: 10,
+						MaxProvisionedThroughput:  100,
+						MinProvisionedConcurrency: 1,
+						MinProvisionedThroughput:  10,
+						Name:                      "entity-name",
+						ProvisionedModelUnits:     100,
+						ExternalModel:             nil,
+						ForceSendFields:           nil,
+					},
+				},
+				TrafficConfig: &serving.TrafficConfig{
+					Routes: []serving.Route{
+						{
+							ServedModelName:   "model-name-1",
+							TrafficPercentage: 100,
+						},
+					},
+				},
+			},
+		},
+	},
 }
 
 type prepareWorkspace func(client *databricks.WorkspaceClient) (any, error)
@@ -253,6 +298,33 @@ var testDeps = map[string]prepareWorkspace{
 		}, nil
 	},
 
+	"model_serving_endpoints.permissions": func(client *databricks.WorkspaceClient) (any, error) {
+		waiter, err := client.ServingEndpoints.Create(context.Background(), serving.CreateServingEndpoint{
+			Name: "endpoint-permissions",
+			Config: &serving.EndpointCoreConfigInput{
+				ServedModels: []serving.ServedModelInput{
+					{
+						ModelName:          "model-name",
+						ModelVersion:       "1",
+						WorkloadSize:       "Small",
+						ScaleToZeroEnabled: true,
+					},
+				},
+			},
+		})
+		if err != nil {
+			return nil, err
+		}
+
+		return &PermissionsState{
+			ObjectID: "/serving-endpoints/" + waiter.Response.Name,
+			Permissions: []iam.AccessControlRequest{{
+				PermissionLevel: "CAN_MANAGE",
+				UserName:        "user@example.com",
+			}},
+		}, nil
+	},
+
 	"schemas.grants": func(client *databricks.WorkspaceClient) (any, error) {
 		return &GrantsState{
 			SecurableType: "schema",
diff --git a/bundle/direct/dresources/model_serving_endpoint.go b/bundle/direct/dresources/model_serving_endpoint.go
new file mode 100644
index 0000000000..a50c635dd2
--- /dev/null
+++ b/bundle/direct/dresources/model_serving_endpoint.go
@@ -0,0 +1,311 @@
+package dresources
+
+import (
+	"context"
+	"fmt"
+	"time"
+
+	"github.com/databricks/cli/bundle/config/resources"
+	"github.com/databricks/cli/bundle/deployplan"
+	"github.com/databricks/cli/libs/utils"
+	"github.com/databricks/databricks-sdk-go"
+	"github.com/databricks/databricks-sdk-go/service/serving"
+	"golang.org/x/sync/errgroup"
+)
+
+type ResourceModelServingEndpoint struct {
+	client *databricks.WorkspaceClient
+}
+
+func (*ResourceModelServingEndpoint) New(client *databricks.WorkspaceClient) *ResourceModelServingEndpoint {
+	return &ResourceModelServingEndpoint{
+		client: client,
+	}
+}
+
+func (*ResourceModelServingEndpoint) PrepareState(input *resources.ModelServingEndpoint) *serving.CreateServingEndpoint {
+	return &input.CreateServingEndpoint
+}
+
+func autoCaptureConfigOutputToInput(output *serving.AutoCaptureConfigOutput) *serving.AutoCaptureConfigInput {
+	if output == nil {
+		return nil
+	}
+	return &serving.AutoCaptureConfigInput{
+		CatalogName:     output.CatalogName,
+		SchemaName:      output.SchemaName,
+		TableNamePrefix: output.TableNamePrefix,
+		Enabled:         output.Enabled,
+		ForceSendFields: utils.FilterFields[serving.AutoCaptureConfigInput](output.ForceSendFields),
+	}
+}
+
+func servedEntitiesOutputToInput(output []serving.ServedEntityOutput) []serving.ServedEntityInput {
+	entities := make([]serving.ServedEntityInput, len(output))
+	for i, entity := range output {
+		entities[i] = serving.ServedEntityInput{
+			EntityName:                entity.EntityName,
+			EntityVersion:             entity.EntityVersion,
+			EnvironmentVars:           entity.EnvironmentVars,
+			ExternalModel:             entity.ExternalModel,
+			InstanceProfileArn:        entity.InstanceProfileArn,
+			MaxProvisionedConcurrency: entity.MaxProvisionedConcurrency,
+			MaxProvisionedThroughput:  entity.MaxProvisionedThroughput,
+			MinProvisionedConcurrency: entity.MinProvisionedConcurrency,
+			MinProvisionedThroughput:  entity.MinProvisionedThroughput,
+			Name:                      entity.Name,
+			ProvisionedModelUnits:     entity.ProvisionedModelUnits,
+			ScaleToZeroEnabled:        entity.ScaleToZeroEnabled,
+			WorkloadSize:              entity.WorkloadSize,
+			WorkloadType:              entity.WorkloadType,
+			ForceSendFields:           utils.FilterFields[serving.ServedEntityInput](entity.ForceSendFields),
+		}
+	}
+
+	return entities
+}
+
+func configOutputToInput(output *serving.EndpointCoreConfigOutput) *serving.EndpointCoreConfigInput {
+	if output == nil {
+		return nil
+	}
+	return &serving.EndpointCoreConfigInput{
+		AutoCaptureConfig: autoCaptureConfigOutputToInput(output.AutoCaptureConfig),
+		ServedEntities:    servedEntitiesOutputToInput(output.ServedEntities),
+		ServedModels:      nil,
+		TrafficConfig:     output.TrafficConfig,
+		Name:              "",
+	}
+}
+
+func (*ResourceModelServingEndpoint) RemapState(state *RefreshOutput) *serving.CreateServingEndpoint {
+	details := state.EndpointDetails
+	// Map the remote state (ServingEndpointDetailed) to the local state (CreateServingEndpoint)
+	// for proper comparison during diff calculation
+	return &serving.CreateServingEndpoint{
+		AiGateway:          details.AiGateway,
+		BudgetPolicyId:     details.BudgetPolicyId,
+		Config:             configOutputToInput(details.Config),
+		Description:        details.Description,
+		EmailNotifications: details.EmailNotifications,
+		Name:               details.Name,
+		RouteOptimized:     details.RouteOptimized,
+		Tags:               details.Tags,
+		ForceSendFields:    utils.FilterFields[serving.CreateServingEndpoint](details.ForceSendFields),
+
+		// Rate limits are a deprecated field that are not returned by the API on GET calls. Thus we map them to nil.
+		// TODO(shreyas): Add a warning when users try setting top level rate limits.
+		RateLimits: nil,
+	}
+}
+
+type RefreshOutput struct {
+	EndpointDetails *serving.ServingEndpointDetailed `json:"endpoint_details"`
+	EndpointId      string                           `json:"endpoint_id"`
+}
+
+func (r *ResourceModelServingEndpoint) DoRefresh(ctx context.Context, id string) (*RefreshOutput, error) {
+	endpoint, err := r.client.ServingEndpoints.GetByName(ctx, id)
+	if err != nil {
+		return nil, err
+	}
+	return &RefreshOutput{
+		EndpointDetails: endpoint,
+		EndpointId:      endpoint.Id,
+	}, nil
+}
+
+func (r *ResourceModelServingEndpoint) DoCreate(ctx context.Context, config *serving.CreateServingEndpoint) (string, error) {
+	waiter, err := r.client.ServingEndpoints.Create(ctx, *config)
+	if err != nil {
+		return "", err
+	}
+
+	return waiter.Response.Name, nil
+}
+
+// waitForEndpointReady waits for the serving endpoint to be ready (not updating)
+func (r *ResourceModelServingEndpoint) waitForEndpointReady(ctx context.Context, name string) (*RefreshOutput, error) {
+	details, err := r.client.ServingEndpoints.WaitGetServingEndpointNotUpdating(ctx, name, 35*time.Minute, nil)
+	if err != nil {
+		return nil, err
+	}
+
+	return &RefreshOutput{
+		EndpointDetails: details,
+		EndpointId:      details.Id,
+	}, nil
+}
+
+func (r *ResourceModelServingEndpoint) WaitAfterCreate(ctx context.Context, config *serving.CreateServingEndpoint) (*RefreshOutput, error) {
+	return r.waitForEndpointReady(ctx, config.Name)
+}
+
+func (r *ResourceModelServingEndpoint) updateAiGateway(ctx context.Context, id string, aiGateway *serving.AiGatewayConfig) error {
+	if aiGateway == nil {
+		req := serving.PutAiGatewayRequest{
+			Name:                 id,
+			FallbackConfig:       nil,
+			Guardrails:           nil,
+			InferenceTableConfig: nil,
+			RateLimits:           nil,
+			UsageTrackingConfig:  nil,
+		}
+		_, err := r.client.ServingEndpoints.PutAiGateway(ctx, req)
+		return err
+	}
+
+	req := serving.PutAiGatewayRequest{
+		Name:                 id,
+		FallbackConfig:       aiGateway.FallbackConfig,
+		Guardrails:           aiGateway.Guardrails,
+		InferenceTableConfig: aiGateway.InferenceTableConfig,
+		RateLimits:           aiGateway.RateLimits,
+		UsageTrackingConfig:  aiGateway.UsageTrackingConfig,
+	}
+	_, err := r.client.ServingEndpoints.PutAiGateway(ctx, req)
+	if err != nil {
+		return fmt.Errorf("failed to update AI gateway: %w", err)
+	}
+	return nil
+}
+
+// TODO(shreyas): Add acceptance tests for update and for unsetting. Will be done once we add selective updates for these fields.
+func (r *ResourceModelServingEndpoint) updateConfig(ctx context.Context, id string, config *serving.EndpointCoreConfigInput) error {
+	if config == nil {
+		// Unset config in resource.
+		req := serving.EndpointCoreConfigInput{
+			Name:              id,
+			AutoCaptureConfig: nil,
+			ServedEntities:    nil,
+			TrafficConfig:     nil,
+			ServedModels:      nil,
+		}
+		_, err := r.client.ServingEndpoints.UpdateConfig(ctx, req)
+		return err
+	}
+	req := serving.EndpointCoreConfigInput{
+		Name:              id,
+		AutoCaptureConfig: config.AutoCaptureConfig,
+		ServedEntities:    config.ServedEntities,
+		TrafficConfig:     config.TrafficConfig,
+		ServedModels:      config.ServedModels,
+	}
+	_, err := r.client.ServingEndpoints.UpdateConfig(ctx, req)
+	if err != nil {
+		return fmt.Errorf("failed to update config: %w", err)
+	}
+	return nil
+}
+
+func (r *ResourceModelServingEndpoint) updateNotifications(ctx context.Context, id string, notifications *serving.EmailNotifications) error {
+	req := serving.UpdateInferenceEndpointNotifications{
+		Name:               id,
+		EmailNotifications: notifications,
+	}
+	_, err := r.client.ServingEndpoints.UpdateNotifications(ctx, req)
+	if err != nil {
+		return fmt.Errorf("failed to update notifications: %w", err)
+	}
+	return nil
+}
+
+func diffTags(currentTags, desiredTags []serving.EndpointTag) (addTags []serving.EndpointTag, deleteTags []string) {
+	addTags = make([]serving.EndpointTag, 0)
+
+	// build maps for easy lookup.
+	currentTagsMap := make(map[string]string)
+	desiredTagsMap := make(map[string]string)
+	for _, tag := range currentTags {
+		currentTagsMap[tag.Key] = tag.Value
+	}
+	for _, tag := range desiredTags {
+		desiredTagsMap[tag.Key] = tag.Value
+	}
+
+	// Compute keys to be added.
+	for key, desiredValue := range desiredTagsMap {
+		v, ok := currentTagsMap[key]
+		if !ok {
+			addTags = append(addTags, serving.EndpointTag{
+				Key:             key,
+				Value:           desiredValue,
+				ForceSendFields: nil,
+			})
+			continue
+		}
+		if v != desiredValue {
+			addTags = append(addTags, serving.EndpointTag{
+				Key:             key,
+				Value:           desiredValue,
+				ForceSendFields: nil,
+			})
+		}
+	}
+
+	// Compute keys to be deleted.
+	for key := range currentTagsMap {
+		if _, ok := desiredTagsMap[key]; !ok {
+			deleteTags = append(deleteTags, key)
+		}
+	}
+
+	return addTags, deleteTags
+}
+
+func (r *ResourceModelServingEndpoint) updateTags(ctx context.Context, id string, tags []serving.EndpointTag) error {
+	endpoint, err := r.client.ServingEndpoints.GetByName(ctx, id)
+	if err != nil {
+		return err
+	}
+
+	addTags, deleteTags := diffTags(endpoint.Tags, tags)
+
+	req := serving.PatchServingEndpointTags{
+		Name:       id,
+		AddTags:    addTags,
+		DeleteTags: deleteTags,
+	}
+	_, err = r.client.ServingEndpoints.Patch(ctx, req)
+	if err != nil {
+		return fmt.Errorf("failed to update tags: %w", err)
+	}
+	return nil
+}
+
+func (r *ResourceModelServingEndpoint) DoUpdate(ctx context.Context, id string, config *serving.CreateServingEndpoint) error {
+	errGroup := errgroup.Group{}
+	errGroup.Go(func() error {
+		return r.updateAiGateway(ctx, id, config.AiGateway)
+	})
+	errGroup.Go(func() error {
+		return r.updateConfig(ctx, id, config.Config)
+	})
+	errGroup.Go(func() error {
+		return r.updateNotifications(ctx, id, config.EmailNotifications)
+	})
+	errGroup.Go(func() error {
+		return r.updateTags(ctx, id, config.Tags)
+	})
+	return errGroup.Wait()
+}
+
+func (r *ResourceModelServingEndpoint) WaitAfterUpdate(ctx context.Context, config *serving.CreateServingEndpoint) (*RefreshOutput, error) {
+	return r.waitForEndpointReady(ctx, config.Name)
+}
+
+func (r *ResourceModelServingEndpoint) DoDelete(ctx context.Context, id string) error {
+	return r.client.ServingEndpoints.DeleteByName(ctx, id)
+}
+
+func (*ResourceModelServingEndpoint) FieldTriggers(_ bool) map[string]deployplan.ActionType {
+	// TF implementation: https://github.com/databricks/terraform-provider-databricks/blob/6c106e8e7052bb2726148d66309fd460ed444236/mlflow/resource_mlflow_experiment.go#L22
+	return map[string]deployplan.ActionType{
+		"name":        deployplan.ActionTypeRecreate,
+		"description": deployplan.ActionTypeRecreate, // description is immutable, can't be updated via API
+		"config.auto_capture_config.catalog_name":      deployplan.ActionTypeRecreate,
+		"config.auto_capture_config.schema_name":       deployplan.ActionTypeRecreate,
+		"config.auto_capture_config.table_name_prefix": deployplan.ActionTypeRecreate,
+		"route_optimized":                              deployplan.ActionTypeRecreate,
+	}
+}
diff --git a/bundle/direct/dresources/permissions.go b/bundle/direct/dresources/permissions.go
index ce06a853a2..b6dbf4c6ad 100644
--- a/bundle/direct/dresources/permissions.go
+++ b/bundle/direct/dresources/permissions.go
@@ -57,13 +57,22 @@ func PreparePermissionsInputConfig(inputConfig any, node string) (*structvar.Str
 		})
 	}
 
+	objectIdRef := prefix + "${" + baseNode + ".id}"
+	// For permissions, model serving endpoint uses it's internal ID, which is different
+	// from its CRUD APIs which use the name.
+	// We have a wrapper struct [RefreshOutput] from which we read the internal ID
+	// in order to set the appropriate permissions.
+	if strings.HasPrefix(baseNode, "resources.model_serving_endpoints.") {
+		objectIdRef = prefix + "${" + baseNode + ".endpoint_id}"
+	}
+
 	return &structvar.StructVar{
 		Config: &PermissionsState{
 			ObjectID:    "", // Always a reference, defined in Refs below
 			Permissions: permissions,
 		},
 		Refs: map[string]string{
-			"object_id": prefix + "${" + baseNode + ".id}",
+			"object_id": objectIdRef,
 		},
 	}, nil
 }
diff --git a/libs/testserver/fake_workspace.go b/libs/testserver/fake_workspace.go
index b1fbf90fd1..89af48778b 100644
--- a/libs/testserver/fake_workspace.go
+++ b/libs/testserver/fake_workspace.go
@@ -22,6 +22,7 @@ import (
 	"github.com/databricks/databricks-sdk-go/service/jobs"
 	"github.com/databricks/databricks-sdk-go/service/ml"
 	"github.com/databricks/databricks-sdk-go/service/pipelines"
+	"github.com/databricks/databricks-sdk-go/service/serving"
 	"github.com/databricks/databricks-sdk-go/service/sql"
 	"github.com/databricks/databricks-sdk-go/service/workspace"
 )
@@ -134,6 +135,7 @@ type FakeWorkspace struct {
 	Clusters            map[string]compute.ClusterDetails
 	Catalogs            map[string]catalog.CatalogInfo
 	RegisteredModels    map[string]catalog.RegisteredModelInfo
+	ServingEndpoints    map[string]serving.ServingEndpointDetailed
 
 	Acls map[string][]workspace.AclItem
 
@@ -232,6 +234,7 @@ func NewFakeWorkspace(url, token string) *FakeWorkspace {
 		Dashboards:           map[string]fakeDashboard{},
 		PublishedDashboards:  map[string]dashboards.PublishedDashboard{},
 		SqlWarehouses:        map[string]sql.GetWarehouseResponse{},
+		ServingEndpoints:     map[string]serving.ServingEndpointDetailed{},
 		Repos:                map[string]workspace.RepoInfo{},
 		Acls:                 map[string][]workspace.AclItem{},
 		Permissions:          map[string]iam.ObjectPermissions{},
diff --git a/libs/testserver/handlers.go b/libs/testserver/handlers.go
index 5c1b6c784a..56ff7616d4 100644
--- a/libs/testserver/handlers.go
+++ b/libs/testserver/handlers.go
@@ -603,6 +603,35 @@ func AddDefaultHandlers(server *Server) {
 		return MapDelete(req.Workspace, req.Workspace.ModelRegistryModels, req.URL.Query().Get("name"))
 	})
 
+	// Serving Endpoints:
+	server.Handle("GET", "/api/2.0/serving-endpoints/{name}", func(req Request) any {
+		return MapGet(req.Workspace, req.Workspace.ServingEndpoints, req.Vars["name"])
+	})
+
+	server.Handle("POST", "/api/2.0/serving-endpoints", func(req Request) any {
+		return req.Workspace.ServingEndpointCreate(req)
+	})
+
+	server.Handle("PUT", "/api/2.0/serving-endpoints/{name}/config", func(req Request) any {
+		return req.Workspace.ServingEndpointUpdate(req, req.Vars["name"])
+	})
+
+	server.Handle("DELETE", "/api/2.0/serving-endpoints/{name}", func(req Request) any {
+		return MapDelete(req.Workspace, req.Workspace.ServingEndpoints, req.Vars["name"])
+	})
+
+	server.Handle("PUT", "/api/2.0/serving-endpoints/{name}/ai-gateway", func(req Request) any {
+		return req.Workspace.ServingEndpointPutAiGateway(req, req.Vars["name"])
+	})
+
+	server.Handle("PATCH", "/api/2.0/serving-endpoints/{name}/notifications", func(req Request) any {
+		return req.Workspace.ServingEndpointUpdateNotifications(req, req.Vars["name"])
+	})
+
+	server.Handle("PATCH", "/api/2.0/serving-endpoints/{name}/tags", func(req Request) any {
+		return req.Workspace.ServingEndpointPatchTags(req, req.Vars["name"])
+	})
+
 	// Generic permissions endpoints
 	server.Handle("GET", "/api/2.0/permissions/{object_type}/{object_id}", func(req Request) any {
 		return req.Workspace.GetPermissions(req)
diff --git a/libs/testserver/serving_endpoints.go b/libs/testserver/serving_endpoints.go
new file mode 100644
index 0000000000..86dfc4babe
--- /dev/null
+++ b/libs/testserver/serving_endpoints.go
@@ -0,0 +1,312 @@
+package testserver
+
+import (
+	"encoding/json"
+	"fmt"
+	"sort"
+
+	"github.com/databricks/databricks-sdk-go/service/serving"
+)
+
+func servedEntitiesInputToOutput(input []serving.ServedEntityInput) []serving.ServedEntityOutput {
+	entities := make([]serving.ServedEntityOutput, len(input))
+	for i, entity := range input {
+		entities[i] = serving.ServedEntityOutput{
+			EntityName:                entity.EntityName,
+			EntityVersion:             entity.EntityVersion,
+			EnvironmentVars:           entity.EnvironmentVars,
+			ExternalModel:             entity.ExternalModel,
+			InstanceProfileArn:        entity.InstanceProfileArn,
+			MaxProvisionedConcurrency: entity.MaxProvisionedConcurrency,
+			MaxProvisionedThroughput:  entity.MaxProvisionedThroughput,
+			MinProvisionedConcurrency: entity.MinProvisionedConcurrency,
+			MinProvisionedThroughput:  entity.MinProvisionedThroughput,
+			Name:                      entity.Name,
+			ProvisionedModelUnits:     entity.ProvisionedModelUnits,
+			ScaleToZeroEnabled:        entity.ScaleToZeroEnabled,
+			WorkloadSize:              entity.WorkloadSize,
+			WorkloadType:              entity.WorkloadType,
+			ForceSendFields:           entity.ForceSendFields,
+		}
+	}
+	return entities
+}
+
+func servedModelsInputToOutput(input []serving.ServedModelInput) []serving.ServedModelOutput {
+	models := make([]serving.ServedModelOutput, len(input))
+	for i, model := range input {
+		models[i] = serving.ServedModelOutput{
+			ModelName:                 model.ModelName,
+			ModelVersion:              model.ModelVersion,
+			EnvironmentVars:           model.EnvironmentVars,
+			InstanceProfileArn:        model.InstanceProfileArn,
+			MaxProvisionedConcurrency: model.MaxProvisionedConcurrency,
+			MinProvisionedConcurrency: model.MinProvisionedConcurrency,
+			ProvisionedModelUnits:     model.ProvisionedModelUnits,
+			ScaleToZeroEnabled:        model.ScaleToZeroEnabled,
+			WorkloadSize:              model.WorkloadSize,
+			WorkloadType:              serving.ServingModelWorkloadType(model.WorkloadType),
+			ForceSendFields:           model.ForceSendFields,
+		}
+	}
+	return models
+}
+
+func autoCaptureConfigInputToOutput(input *serving.AutoCaptureConfigInput) *serving.AutoCaptureConfigOutput {
+	return &serving.AutoCaptureConfigOutput{
+		CatalogName:     input.CatalogName,
+		SchemaName:      input.SchemaName,
+		TableNamePrefix: input.TableNamePrefix,
+		Enabled:         input.Enabled,
+		ForceSendFields: input.ForceSendFields,
+	}
+}
+
+func (s *FakeWorkspace) ServingEndpointCreate(req Request) Response {
+	defer s.LockUnlock()()
+
+	var createReq serving.CreateServingEndpoint
+	err := json.Unmarshal(req.Body, &createReq)
+	if err != nil {
+		return Response{
+			Body:       fmt.Sprintf("cannot unmarshal request body: %s", err),
+			StatusCode: 400,
+		}
+	}
+
+	// Check if endpoint with this name already exists
+	if _, exists := s.ServingEndpoints[createReq.Name]; exists {
+		return Response{
+			StatusCode: 409,
+			Body:       map[string]string{"error_code": "RESOURCE_ALREADY_EXISTS", "message": fmt.Sprintf("Serving endpoint with name %s already exists", createReq.Name)},
+		}
+	}
+
+	// Convert config to output format
+	var config *serving.EndpointCoreConfigOutput
+	if createReq.Config != nil {
+		config = &serving.EndpointCoreConfigOutput{
+			TrafficConfig: createReq.Config.TrafficConfig,
+		}
+
+		// Convert ServedEntityInput to ServedEntityOutput
+		if len(createReq.Config.ServedEntities) > 0 {
+			config.ServedEntities = servedEntitiesInputToOutput(createReq.Config.ServedEntities)
+		}
+
+		// Convert ServedModelInput to ServedModelOutput
+		if len(createReq.Config.ServedModels) > 0 {
+			config.ServedModels = servedModelsInputToOutput(createReq.Config.ServedModels)
+		}
+
+		// Convert AutoCaptureConfig if present
+		if createReq.Config.AutoCaptureConfig != nil {
+			config.AutoCaptureConfig = autoCaptureConfigInputToOutput(createReq.Config.AutoCaptureConfig)
+		}
+	}
+
+	endpoint := serving.ServingEndpointDetailed{
+		AiGateway:          createReq.AiGateway,
+		BudgetPolicyId:     createReq.BudgetPolicyId,
+		Config:             config,
+		Creator:            s.CurrentUser().UserName,
+		Description:        createReq.Description,
+		EmailNotifications: createReq.EmailNotifications,
+		Id:                 nextUUID(),
+		Name:               createReq.Name,
+		RouteOptimized:     createReq.RouteOptimized,
+		Tags:               createReq.Tags,
+		State: &serving.EndpointState{
+			ConfigUpdate: serving.EndpointStateConfigUpdateNotUpdating,
+		},
+		ForceSendFields: createReq.ForceSendFields,
+	}
+
+	s.ServingEndpoints[createReq.Name] = endpoint
+
+	return Response{
+		Body: endpoint,
+	}
+}
+
+func (s *FakeWorkspace) ServingEndpointUpdate(req Request, name string) Response {
+	defer s.LockUnlock()()
+
+	var updateReq serving.EndpointCoreConfigInput
+	err := json.Unmarshal(req.Body, &updateReq)
+	if err != nil {
+		return Response{
+			Body:       fmt.Sprintf("cannot unmarshal request body: %s", err),
+			StatusCode: 400,
+		}
+	}
+
+	endpoint, exists := s.ServingEndpoints[name]
+	if !exists {
+		return Response{
+			StatusCode: 404,
+			Body:       map[string]string{"error_code": "RESOURCE_DOES_NOT_EXIST", "message": fmt.Sprintf("Serving endpoint with name %s not found", name)},
+		}
+	}
+
+	// Convert config to output format
+	var config *serving.EndpointCoreConfigOutput
+	if updateReq.ServedEntities != nil || updateReq.ServedModels != nil || updateReq.TrafficConfig != nil {
+		config = &serving.EndpointCoreConfigOutput{
+			TrafficConfig: updateReq.TrafficConfig,
+		}
+
+		// Convert ServedEntityInput to ServedEntityOutput
+		if len(updateReq.ServedEntities) > 0 {
+			config.ServedEntities = servedEntitiesInputToOutput(updateReq.ServedEntities)
+		}
+
+		// Convert ServedModelInput to ServedModelOutput
+		if len(updateReq.ServedModels) > 0 {
+			config.ServedModels = servedModelsInputToOutput(updateReq.ServedModels)
+		}
+
+		// Convert AutoCaptureConfig if present
+		if updateReq.AutoCaptureConfig != nil {
+			config.AutoCaptureConfig = autoCaptureConfigInputToOutput(updateReq.AutoCaptureConfig)
+		}
+	}
+
+	endpoint.Config = config
+	endpoint.State = &serving.EndpointState{
+		ConfigUpdate: serving.EndpointStateConfigUpdateNotUpdating,
+	}
+
+	s.ServingEndpoints[name] = endpoint
+
+	return Response{
+		Body: endpoint,
+	}
+}
+
+func (s *FakeWorkspace) ServingEndpointPutAiGateway(req Request, name string) Response {
+	defer s.LockUnlock()()
+
+	var putReq serving.PutAiGatewayRequest
+	err := json.Unmarshal(req.Body, &putReq)
+	if err != nil {
+		return Response{
+			Body:       fmt.Sprintf("cannot unmarshal request body: %s", err),
+			StatusCode: 400,
+		}
+	}
+
+	endpoint, exists := s.ServingEndpoints[name]
+	if !exists {
+		return Response{
+			StatusCode: 404,
+			Body:       map[string]string{"error_code": "RESOURCE_DOES_NOT_EXIST", "message": fmt.Sprintf("Serving endpoint with name %s not found", name)},
+		}
+	}
+
+	// Update AI gateway config
+	if putReq.FallbackConfig != nil || putReq.Guardrails != nil || putReq.InferenceTableConfig != nil || putReq.RateLimits != nil || putReq.UsageTrackingConfig != nil {
+		endpoint.AiGateway = &serving.AiGatewayConfig{
+			FallbackConfig:       putReq.FallbackConfig,
+			Guardrails:           putReq.Guardrails,
+			InferenceTableConfig: putReq.InferenceTableConfig,
+			RateLimits:           putReq.RateLimits,
+			UsageTrackingConfig:  putReq.UsageTrackingConfig,
+		}
+	} else {
+		// Unset AI gateway if no fields provided
+		endpoint.AiGateway = nil
+	}
+
+	s.ServingEndpoints[name] = endpoint
+
+	return Response{
+		Body: endpoint.AiGateway,
+	}
+}
+
+func (s *FakeWorkspace) ServingEndpointUpdateNotifications(req Request, name string) Response {
+	defer s.LockUnlock()()
+
+	var updateReq serving.UpdateInferenceEndpointNotifications
+	err := json.Unmarshal(req.Body, &updateReq)
+	if err != nil {
+		return Response{
+			Body:       fmt.Sprintf("cannot unmarshal request body: %s", err),
+			StatusCode: 400,
+		}
+	}
+
+	endpoint, exists := s.ServingEndpoints[name]
+	if !exists {
+		return Response{
+			StatusCode: 404,
+			Body:       map[string]string{"error_code": "RESOURCE_DOES_NOT_EXIST", "message": fmt.Sprintf("Serving endpoint with name %s not found", name)},
+		}
+	}
+
+	endpoint.EmailNotifications = updateReq.EmailNotifications
+	s.ServingEndpoints[name] = endpoint
+
+	return Response{
+		Body: endpoint,
+	}
+}
+
+func (s *FakeWorkspace) ServingEndpointPatchTags(req Request, name string) Response {
+	defer s.LockUnlock()()
+
+	var patchReq serving.PatchServingEndpointTags
+	err := json.Unmarshal(req.Body, &patchReq)
+	if err != nil {
+		return Response{
+			Body:       fmt.Sprintf("cannot unmarshal request body: %s", err),
+			StatusCode: 400,
+		}
+	}
+
+	endpoint, exists := s.ServingEndpoints[name]
+	if !exists {
+		return Response{
+			StatusCode: 404,
+			Body:       map[string]string{"error_code": "RESOURCE_DOES_NOT_EXIST", "message": fmt.Sprintf("Serving endpoint with name %s not found", name)},
+		}
+	}
+
+	// Build map of current tags
+	tagMap := make(map[string]string)
+	for _, tag := range endpoint.Tags {
+		tagMap[tag.Key] = tag.Value
+	}
+
+	// Add or update tags
+	for _, tag := range patchReq.AddTags {
+		tagMap[tag.Key] = tag.Value
+	}
+
+	// Delete tags
+	for _, key := range patchReq.DeleteTags {
+		delete(tagMap, key)
+	}
+
+	// Convert back to slice sorted by key for stable output
+	tags := make([]serving.EndpointTag, 0, len(tagMap))
+	keys := make([]string, 0, len(tagMap))
+	for key := range tagMap {
+		keys = append(keys, key)
+	}
+	sort.Strings(keys)
+	for _, key := range keys {
+		tags = append(tags, serving.EndpointTag{Key: key, Value: tagMap[key]})
+	}
+
+	endpoint.Tags = tags
+	s.ServingEndpoints[name] = endpoint
+
+	// Return the tags as EndpointTags struct, not as array
+	return Response{
+		Body: serving.EndpointTags{
+			Tags: tags,
+		},
+	}
+}
diff --git a/out.first-get..json b/out.first-get..json
new file mode 100644
index 0000000000..e69de29bb2