Snowflake AI pricing¶
Snowflake AI features use AI Credit pricing, a flat, simplified pricing model for AI services. AI Credits are separate from Platform Credits and provide consistent pricing regardless of your Snowflake edition. AI costs scale with your usage. Each service is charged per the relevant table in the Consumption table. There are no per-seat fees. Automatic AI Credit discounts apply based on your annual contract value (ACV).
AI Credits compared to Platform Credits¶
Snowflake compute costs are quoted in credits. There are two types:
-
AI Credits: Apply to AI features listed in Table 6 of Snowflake’s Consumption Table. AI Credits provide a flat pricing model regardless of edition (Standard, Enterprise, Business Critical, or VPS) or region. AI Credits have the same price regardless of customer edition.
-
Platform Credits: The default credit price for all other Snowflake usage. Any pricing not specifically identified as AI Credit pricing uses Platform Credits. Platform Credit prices may vary by edition and region.
AI Credit price¶
| Routing type | Price per AI Credit | When it applies |
|---|---|---|
| Global routing | $2.00 | CORTEX_ENABLED_CROSS_REGION is set to ANY_REGION, AWS_GLOBAL, GCP_GLOBAL, or AZURE_GLOBAL. Requests can be routed across broader Snowflake-supported capacity, providing the lowest AI Credit price and broader model availability. |
| Regional routing | $2.20 | CORTEX_ENABLED_CROSS_REGION is set to DISABLED or a specific regional setting such as AWS_US, AWS_EU, AZURE_US, or AZURE_EU. Requests are restricted to the account’s home region or selected geographies, typically for data residency or compliance requirements. |
For more information about the cross-region routing parameter, see CORTEX_ENABLED_CROSS_REGION.
Features that use AI Credit pricing¶
The following features use AI Credit pricing. For complete details, see Table 6 of Snowflake’s Consumption Table.
| Feature | Pricing type |
|---|---|
| Cortex Code | AI Credit |
| Cortex Agents | AI Credit |
| Snowflake Intelligence | AI Credit |
| Cortex REST API | AI Credit |
| AI Functions | AI Credit |
| Cortex Search | AI Credit |
| Cortex Batch Search | AI Credit |
| AI Parse Doc | AI Credit |
| Cortex Fine-tuning | Platform Credit (legacy) |
| Cortex Analyst API | Platform Credit (legacy) |
Feature pricing details¶
Snowflake Intelligence¶
Snowflake Intelligence is billed in AI Credits based on token consumption. Snowflake Intelligence uses agents to orchestrate multi-step workflows. Costs are additive across the underlying services the agent invokes (such as Cortex Analyst and Cortex Search). The specific rate depends on the model selected for orchestration. You can monitor per-agent usage through the SNOWFLAKE_INTELLIGENCE_USAGE_HISTORY view.
Cortex Agents¶
Cortex Agents are billed in AI Credits per million tokens processed. Because agents orchestrate multi-step workflows, costs are additive across the underlying services the agent invokes (such as Cortex Analyst and Cortex Search). The specific rate depends on the model selected for orchestration. You can monitor per-agent usage through the CORTEX_AGENT_USAGE_HISTORY view.
Cortex Analyst¶
Snowflake recommends invoking Cortex Analyst through Cortex Agents, in which case the pricing is the same token-based AI Credit model as Cortex Agents. If you invoke Cortex Analyst directly through the standalone Analyst API, it’s billed per 1,000 messages. In both cases, executing the generated SQL incurs standard virtual warehouse compute charges.
Cortex Search¶
Cortex Search is billed based on two components: serving compute and embedding compute. Serving compute is charged Credits per GB per month of indexed data (including vector embeddings) and runs continuously while the service is resumed. Embedding compute is charged per token when data is inserted or updated. For more details, see Understanding cost for Cortex Search Services.
Cortex Batch Search¶
Cortex Batch Search is billed Credits per GB per hour of indexed data for the duration of the batch job. Unlike interactive Cortex Search, serving compute only runs while the batch job is active rather than continuously. Additional costs include query embedding tokens and virtual warehouse compute for orchestration.
AI Functions¶
AI Functions (such as AI_COMPLETE, AI_EMBED, AI_CLASSIFY, and AI_EXTRACT) are billed in AI Credits per million tokens processed. The rate varies by model. Both input and output tokens count toward consumption, and you can track usage through the CORTEX_FUNCTIONS_QUERY_USAGE_HISTORY view.
Cortex REST API¶
The Cortex REST API is billed in AI Credits per million tokens. Prompt caching is supported, with cache reads billed at a reduced rate compared to standard input tokens. The specific per-token rate depends on the model you select.
AI Parse Doc¶
AI Parse Doc is billed in AI Credits per 1,000 pages processed. The rate depends on the parsing mode.
Additional details¶
AI Credit pricing only applies to the AI features listed above. Non-AI services such as warehouses, storage, and data transfer continue to use Platform Credit pricing. Your existing capacity balance and contract terms are unchanged, and no contract amendments are required.
No action is required to receive the ACV based discounts. Note that capacity discounts don’t apply to AI Credits.
Example¶
Assume a customer has an Enterprise Edition account, where Platform Credits are $3.00 per credit and AI Credits are $2.00 per credit for global routing. If the customer uses 100 Platform Credits and 100 AI Credits, their spend would be:
- Platform usage: 100 credits × $3.00 = $300
- AI usage: 100 credits × $2.00 = $200
- Total spend: $500