Gemini 4 Argon hits Public Preview in Snowflake Cortex AI

Snowflake frames the addition of Gemini 4 Argon as a response to a mismatch it says it hears from enterprise leaders: no single AI model fits every business challenge. The company's answer is an open, multi-model strategy in which customers pick the right model for each job, and this release is the latest model added to that lineup. The stated goal is to avoid the trade-off between flexibility and control — new models without new data security risks or pipeline complexity.

The announcement sketches the kinds of work it expects this to enable on governed data. Snowflake describes multi-step investigations such as tracing financial discrepancies across billing tables, or using CoCo to refactor legacy code and trace multi-file dependencies. It also points to workflows where visual evidence matters, such as reviewing financial statements, multi-region sales charts and campaign dashboards together in CoWork, and to batch work where operations, legal and support teams apply reasoning models across thousands of unstructured records in parallel.

As an illustration of the underlying pattern rather than this specific model, Snowflake cites Firework, a video commerce company that used Snowpark and Cortex AI to aggregate, clean and classify thousands of anonymous customer conversations, then powered an AI Video Assistant that answers return-policy questions, summarizes product reviews and offers personalized recommendations. Firework reported a simpler process and greater time and cost efficiency by keeping the workflow within Snowflake.