Snowflake Dynamic Model Routing takes aim at AI costs

Enterprise AI is entering a more demanding phase. Companies are no longer just testing models. They are using multiple models across AI apps and agents, making one question harder to ignore: does every task really need the most expensive model?

Snowflake is taking that question directly to its Cortex AI Gateway with Snowflake Dynamic Model Routing. The new capability automatically selects a model based on quality, speed, customer preferences and cost, while giving enterprises more control over AI consumption.

Making model choice less manual

Snowflake says Cortex AI Gateway can route simpler or repetitive tasks to more efficient models, while sending work that needs deeper reasoning to frontier models. The capability is also integrated into Snowflake CoCo and Snowflake CoWork and is available to third-party AI agents using Cortex AI Gateway.

The idea is simple. Instead of development teams deciding which model should handle every request, the gateway makes that decision automatically.

That matters as model choices continue to expand. Snowflake is also adding DeepSeek-V4-Flash 0731 and GLM-5.3 to Snowflake Cortex AI, alongside models from providers including Anthropic, OpenAI, Google, SpaceXAI, Meta and Mistral.

The bigger focus: AI economics

The announcement puts AI cost optimisation at the centre of the model-routing discussion. Snowflake says its internal testing found that combining open and proprietary models can maintain comparable quality while improving token efficiency.

In one evaluation, agents using dynamic model routing built a dbt pipeline with up to 3x greater token efficiency than a frontier-model-only approach while maintaining the same quality. Another test reported 25% greater token efficiency for engineering teams completing the same number of pull requests.

These are Snowflake's internal test results, but they show the thinking behind the feature: model choice is becoming part of the economics of enterprise AI.

More control over AI consumption

Cortex AI Gateway also gives administrators visibility into token usage and costs. Organisations can establish spending limits across AI apps and agents, while Snowflake CoCo adds controls for default models, team or cost-centre attribution, per-user quotas and consumption alerts.

The combination of model routing, AI cost optimisation and usage controls gives enterprises a way to manage AI consumption as adoption grows.

Read More: 

AWS Builder expands to Hyderabad with permanent tech hub

Xebia and OpenAI partnership takes enterprise AI further

Beyond implementation: How AI could reshape Liferay partner revenue

Rapoo appoints Trustech AV Solutions to push enterprise collaboration