Gemini 4 Argon arrives in public preview on Snowflake Cortex AI

Where Gemini 4 Argon is available

Gemini 4 Argon is available at launch across Cortex AI Functions. Snowflake says support in Snowflake CoWork, Snowflake CoCo and Cortex Agents is coming soon across supported clouds via cross-region inference.

Snowflake frames the move as part of a multi-model approach. The company says enterprise leaders keep telling it that their competitive advantage lives in their data, but that no single AI model fits every business challenge. In its view, organizations need the flexibility to pick the right model for each job without creating data security risks or pipeline complexity.

Use cases Snowflake describes

The company lists four ways customers can apply the model on data already stored in Snowflake:

• Sustained reasoning across complex workflows: teams can run multi-step investigations, such as tracing financial discrepancies across complex billing tables, or use CoCo to refactor legacy codebases and trace multi-file dependencies.
• Visual evidence in business analysis: in CoWork, teams can evaluate financial statements, multi-region sales charts and campaign performance dashboards at the same time, tying the analysis to chart evidence to spot trend outliers faster.
• Batch document synthesis: through model-selectable AI Functions, operations, legal and support teams can apply reasoning models across thousands of unstructured records, legal filings and support tickets in parallel to extract context and generate detailed summaries.
• Unified governance: because the model runs natively inside Cortex AI, every call stays within Snowflake's security and governance perimeter, which Snowflake presents as an answer to "shadow AI".

Firework as a customer example

Snowflake points to Firework, a video commerce company, as an example of building on this data foundation. Firework wanted to bring the personalized attention of an in-store associate to online shopping. Using Snowpark and Cortex AI, it aggregated, cleaned and classified thousands of anonymous customer conversations to understand consumer interests and pain points.

That work helped power Firework's AI Video Assistant, which answers return-policy questions, summarizes product reviews and offers personalized recommendations. Firework reported a simpler process and greater time and cost efficiency from keeping the end-to-end workflow within Snowflake. Note that Firework's example is not tied to Gemini 4 Argon itself. Snowflake uses it to illustrate the data pipelines into which teams can now plug the new model.

What it means for Snowflake customers

For Snowflake customers, the change adds another model option inside the platform they already use for governed data. Snowflake says teams can plug Google Cloud's latest reasoning and visual capabilities into existing data pipelines to take on more complex, multi-step business challenges.

The release is a Public Preview, and the availability in CoWork, CoCo and Cortex Agents is still to come, so the full range of use cases described depends on those integrations arriving. Snowflake directs interested customers to its Google Cloud Partner Hub to explore its multi-model capabilities. The source gives no details on pricing or regional availability.