Snowflake's Cortex AI Gateway routes AI workloads by cost and speed
Snowflake says its Cortex AI Gateway can automatically pick the best model for a task and apply governance controls to open and proprietary models alike.
Snowflake says capabilities in its Cortex AI Gateway automatically direct AI workloads to the model that best balances quality, speed and cost for a given use case, according to techradar.com. The company describes the model-selection features as recently launched.
The gateway also tracks AI spend, and the routing can weigh quality, speed and customer preferences when choosing among available models. Customers can control which models and providers are made available to their users.
Snowflake says governance policies and controls, including masking, apply to any model handling a request, whether that model is open or proprietary.
Model choice meets spend control
Grace Adamson, AI Product Marketing Manager at Snowflake, discussed the capabilities in an interview with techradar.com.
The company says the first wave of generative AI centred heavily on model power, with experimentation aimed at the highest-quality outputs or the most complex tasks.
Quick answers
What does Snowflake's Cortex AI Gateway do?
It automatically routes AI workloads to the model that best balances quality, speed and cost for a use case, and tracks AI spend, according to Snowflake.
Can customers limit which AI models their users can access?
Yes. Snowflake says customers control which models and providers are available to their users, and that governance policies such as masking apply to any model handling a request.