SnowPro Specialty: Gen AI Question 256
Single answerSnowflake Service Consumption TableA GenAI platform team uses Snowflake Cortex AI functions across several applications and wants to charge back usage to internal teams. They need a reliable way to analyze token-based consumption and credit usage for AI services over time using account telemetry rather than application logs. Which Snowflake object should they query to get service-level consumption details for AI-related usage such as Cortex services?
- A
The SERVICE_CONSUMPTION_TABLE in the ORGANIZATION_USAGE schema
- B
The CORTEX_FUNCTIONS view in ACCOUNT_USAGE
- C
The QUERY_HISTORY view in ACCOUNT_USAGE, because it contains all AI service token usage and charge details
- D
The SERVICE_CONSUMPTION_TABLE in the ACCOUNT_USAGE schema
Show answer and explanation
Correct answer: D
Explanation
For SnowPro Specialty: Gen AI, candidates should understand that operational and chargeback reporting for Snowflake-managed AI services should be based on Snowflake usage telemetry rather than inferred solely from application or query logs. The ACCOUNT_USAGE.SERVICE_CONSUMPTION_TABLE is intended to expose service-level consumption data that can be used to analyze AI-related usage patterns and costs. QUERY_HISTORY may still be useful for correlating usage back to specific SQL statements or workloads, but it is not the primary accounting source for service consumption. In practice, teams often combine SERVICE_CONSUMPTION_TABLE with query history, tags, or internal metadata to implement departmental showback or chargeback. This aligns with Snowflake best practices of using ACCOUNT_USAGE views for account-level observability and governance.
- A. Incorrect.
Incorrect. ORGANIZATION_USAGE provides organization-level usage and billing visibility across accounts, but the Snowflake Service Consumption Table referenced for service-level telemetry is not queried from ORGANIZATION_USAGE for this use case. A team analyzing AI service consumption within a specific account would use the account-level usage view/table intended for service consumption reporting.
- B. Incorrect.
Incorrect. There is no standard ACCOUNT_USAGE view named CORTEX_FUNCTIONS that serves as the authoritative billing and service-consumption source for Cortex AI usage. This distractor targets the misconception that every Snowflake feature has a dedicated usage view named after the feature.
- C. Incorrect.
Incorrect. QUERY_HISTORY can help identify which statements invoked AI functions, but it is not the authoritative source for aggregated service-consumption accounting such as service-level credits and usage telemetry. Relying on query text or history alone for chargeback is incomplete and error-prone.
- D. Correct.
Correct. The SERVICE_CONSUMPTION_TABLE in ACCOUNT_USAGE is the appropriate object to query for account-level service consumption details, including Snowflake-managed service usage such as AI services. It is designed for telemetry and cost analysis scenarios where teams need to monitor and allocate consumption over time.