Comprehensive and Detailed 150 to 250 words of Explanation From [SnowPro® Specialty: Gen AI/Course Guide/topics]:
The Snowflake Model Registry provides unified, centralized machine learning lifecycle management by treating models as first-class, schema-level objects in the Data Cloud.
Two fundamental capabilities define its operational architecture:
1. **Native SQL Inference Deployment:** Once a machine learning or deep learning model is logged and registered, its inference methods can be invoked directly inside standard SQL queries using the syntax ` < model_name > ! < method_name > ( < arguments > )`. This eliminates the need for external serving infrastructure or external microservices, executing inference workloads natively within Snowflake virtual warehouses or container pools.
2. **Native Artifact Versioning:** The Model Registry natively supports comprehensive versioning under a common parent model identifier. Engineering teams can deploy, track, compare, and audit multiple iterations of a model (e.g., v1, v2) side-by-side, designating a default operational version while maintaining full traceability and governance across release cycles.
Option A is incorrect because models and versions support standard lifecycle management, including metadata modification and deletion via SQL (`DROP MODEL`). Option B is false because models are logged directly to internal stages without requiring external storage buckets. Option D is incorrect because registered models can be invoked via SQL as well as Python.
References & Study Guide Topics:
* Snowflake Gen AI Governance: Snowflake Model Registry Architecture & Governance
* MLOps Lifecycle: Schema-Level Model Management, Versioning, and SQL Scoring
===============