SnowPro Specialty: Gen AI exam dumps

SnowPro Specialty: Gen AI practice question 221 of 287

SnowPro® Specialty: Gen AI. Expert level, Snowflake. Free question with the correct answer and a full explanation.

SnowPro Specialty: Gen AI Question 221

Single answerCORTEX_MODELS_ALLOWLIST parameter

A Snowflake administrator is preparing a production environment where analysts can use Snowflake Cortex LLM functions, but the security team requires that only a small set of approved models be available. The admin wants to enforce this restriction centrally so that attempts to use non-approved models are blocked. Which action should the administrator take?

  1. A

    Set the CORTEX_MODELS_ALLOWLIST account parameter to the approved model names so only those Cortex models can be used

  2. B

    Grant USAGE on the approved models to analyst roles and Snowflake will automatically block all other Cortex models

  3. C

    Create a network policy that lists the approved models and attach it to the analyst users

  4. D

    Set CORTEX_MODELS_ALLOWLIST at the warehouse level so each virtual warehouse enforces its own approved model list

Show answer and explanation

Correct answer: A

Explanation

The key requirement in this scenario is to centrally restrict Snowflake Cortex model usage to a defined set of approved models. The purpose of the CORTEX_MODELS_ALLOWLIST parameter is exactly that: it lets administrators limit which Cortex models are available in the environment. This is useful for governance, compliance, and standardization. Common wrong approaches include trying to manage models through object privileges, network policies, or warehouse settings, but those mechanisms do not provide model-level allowlisting. In Snowflake documentation and best-practice guidance for Cortex governance, administrators should use the relevant Cortex model allowlist parameter to enforce which models users can access.

  • A. Correct.

    Correct. CORTEX_MODELS_ALLOWLIST is the parameter used to restrict which Snowflake Cortex models are available for use. Setting it to the approved model names is the appropriate centralized control when an organization wants to limit model usage for governance, compliance, or cost-control reasons.

  • B. Incorrect.

    Incorrect. Cortex model access is not controlled by granting USAGE on individual models in the same way database objects are managed. This reflects a common misconception that models behave like schemas, tables, or stages for privilege assignment. The allowlist parameter is the intended mechanism for restricting available models.

  • C. Incorrect.

    Incorrect. Network policies control client network access to Snowflake, such as allowed IP ranges, not which Cortex models users can invoke. Someone might choose this option because both are security controls, but network policies do not govern model selection.

  • D. Incorrect.

    Incorrect. CORTEX_MODELS_ALLOWLIST is not a warehouse-level control used to scope model availability per virtual warehouse. The misconception here is assuming that AI model controls are tied to compute resources. In practice, this parameter is used as an administrative configuration rather than a warehouse-specific execution setting.

Timed practice exam

Take a SnowPro Specialty: Gen AI practice test under exam conditions

55 questions in 85 minutes, drawn from this bank, with a score report and a per-question review when you finish.

Start timed exam