Snowflake says dynamic model routing can match AI tasks to lower-cost models
Reported by TechRadar1 source ↓
Snowflake says its Cortex AI Gateway can route requests to models based on quality, speed and cost, while applying the same governance controls across models.
Key points
- Snowflake says simple tasks can go to efficient models while complex requests use frontier models
- Customers can control which models and providers are available to users
- Snowflake reports internal tests found up to three times greater token efficiency with routing, at the same quality
Details
In an interview with TechRadar, Grace Adamson, Snowflake’s AI Product Marketing Manager, described dynamic model routing in Cortex AI Gateway. Snowflake says the feature directs each workload to a suitable model according to its quality, speed and cost, rather than using the same model for every request.
Adamson said governance policies, including masking, apply across models, and customers can limit which models and providers are available. Snowflake also said internal tests of agents building a dbt pipeline found up to three times greater token efficiency with dynamic routing than with a frontier-model-only approach, while maintaining the same quality; the article does not provide further test details.
Written from coverage as of
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh