ReleaseBytes
snowflake Snowflake Blog ·

Snowflake Cortex AI Adds Dynamic Model Routing and Expanded Open Model Portfolio

blogaisnowflakepreviewengineer
feature announcement

Snowflake Cortex AI is introducing dynamic model routing and expanding its selection of open models to enhance AI economics and intelligence efficiency for organizations. These new capabilities aim to match AI workloads with the most appropriate model, optimizing both cost and performance. Dynamic model routing, coming to private preview, will intelligently direct requests through the Cortex AI Gateway to reduce inference spend and complexity. The expanded open model portfolio includes DeepSeek-V4-Flash (private preview) and GLM-5.3 (coming soon), served securely within Snowflake's governance boundary.

  • →Dynamic Model Routing in Cortex AI Gateway
  • →Expanded Open Model Portfolio in Cortex AI
Features (2) ›
  • Dynamic Model Routing in Cortex AI Gateway

    Snowflake Cortex AI Gateway will introduce dynamic model routing, a feature designed to automatically select the most cost-effective model capable of completing a task. This capability, coming to private preview, reduces inference costs by directing lower-complexity tasks to efficient models and advanced workloads to frontier models, adapting without requiring application rebuilds or manual logic.

  • Expanded Open Model Portfolio in Cortex AI

    Snowflake is expanding its portfolio of open models within Cortex AI, adding DeepSeek-V4-Flash (private preview) and GLM-5.3 (coming soon). These models are served directly by Snowflake, ensuring inference occurs within a secure perimeter, reducing latency, and adhering to enterprise governance controls and data residency requirements.

Read the original announcement →

https://www.snowflake.com/content/snowflake-site/global/en/blog/dynamic-model-routing-open-models-cortex-ai

Related releases