[REVISION HISTORY]

Snowflake Cortex AI Gateway model routing updates

Updated 1 time since CLSTR started tracking revisions of this situation.

What changed

2026-08-21 02:50 UTC → 2026-09-01 21:29 UTC · added ~~removed~~

Snowflake is implementing dynamic model routing within its Cortex AI Gateway, Snowflake CoCo, and Snowflake CoWork to optimize enterprise AI costs and performance. The system automatically directs routine or low-complexity ~~tasks to~~ tasks—such as classification and data extraction—to efficient, lower-cost models, while reserving frontier models for complex reasoning tasks. This initiative aims to improve ‘Intelligence Efficiency’ by selecting models based on quality, speed, and cost requirements. Internal testing reported efficiency gains of up to three times, including a 25% increase in token efficiency for engineering pull requests and significant improvements for dbt pipeline creation compared to using frontier models exclusively. The routing feature is also being made available to third-party AI agents via the Cortex AI Gateway. As part of this rollout, Snowflake is expanding its library of open-source models, including the addition of GLM-5.3 and the private preview of DeepSeek-V4-Flash 0731. ~~The~~ To support enterprise governance, the company is ~~also~~ introducing ~~new~~ tools ~~for enterprises~~ to track AI usage, allocate costs, and manage spending ~~limits~~ limits. This includes the Yuki Cortex Control Center, a dashboard designed to address the ‘black box’ problem of AI spending by providing real-time visibility into expenditures broken down by product, agent, user, role, and ~~quotas across various teams~~ model, allowing users to evaluate spending alongside cost and ~~agents.~~ latency performance.

Versions

Only revisions since CLSTR began indexing content versions appear here. Select a version to see what changed compared to the one before it.