Cortex Code Costs: Credits, Tokens & Spend Limits

Everyone tells you to use AI, but nobody explains how you get charged for it. This chapter breaks down exactly what Cortex Code costs: AI credits, compute credits, and storage, plus why resource monitors do not cover AI credits. We compare token usage between Snowsight and the CLI, decode input, output, and cached tokens, and explain why output tokens cost far more than input. What you will learn: The three things Snowflake charges for and how credits map to dollars Reading token and credit usage from ACCOUNT_USAGE (with its data lag) Why output tokens run 3 to 4 times costlier than input tokens How prompt caching makes longer sessions cheaper Governance: setting daily estimated credit limits per user for CLI and Snowsight A practical guide to using Cortex Code responsibly without a surprise bill. #Snowflake #CortexCode #FinOps #AICost #Governance