Snowflake Inc. unveiled new AI optimization features designed to cut costs and improve efficiency across its platform, including dynamic model routing and expanded model support.
The company introduced dynamic model routing, which automatically assigns tasks to the most cost-effective AI models based on complexity. Simpler tasks are routed to efficient models, while complex workloads are directed to advanced options. This feature is integrated into Snowflake CoCo and Snowflake CoWork, and is also available to third-party AI agents via the Cortex AI Gateway.
Snowflake also added support for additional AI models, including DeepSeek-V4-Flash 0731, GLM-5.3 and GLM-5.2, through its Cortex AI Gateway. The DeepSeek-V4-Flash 0731 model is currently in private preview, while GLM-5.3 is expected to enter private preview pending model availability. The company’s internal testing showed that agents using dynamic routing built a dbt pipeline with up to three times greater token efficiency compared with using only advanced models, while maintaining output quality.
In a separate evaluation, engineering teams completed the same number of pull requests with 25% greater token efficiency. Snowflake’s AI research team also reported that DeepSeek-V4-Flash scored 74.4% on enterprise data engineering tasks, outperforming GLM-5.2, which achieved 62.8% while consuming fewer tokens.
Administrative controls have been enhanced to support cost management. The Cortex AI Gateway provides visibility into token usage and spending limits across AI applications and agents. Snowflake CoCo extends these capabilities with role-based access, allowing administrators to set default models, allocate usage by team, enforce per-user quotas and receive consumption notifications.
Snowflake serves more than 13,900 customers globally. The new features were announced in July 2026 alongside the launch of the Cortex AI Gateway.
