Bring Your Own Snowflake LLM

Last modified: May 28, 2026

Introduction

You can leverage a LLM (Large Language Model) from your Snowflake account in your Mendix application for GenAI functionality. In Snowflake this is known as BYOK (Bring your own Key). For a list of available LLM models, refer to the Snowflake documentation.

Using a Snowflake BYOK has the following advantages:

  • Easy - Offers unified integration with model provider capacity and multi-cloud resilience
  • Trusted - Leverage the Snowflake Secure Data Boundary
  • Cost-efficient - Benefit from unified billing and Snowflake committed spend draw down

Typical Use Cases

  • Chat interfaces in natural language
  • Usage by Mendix AI Agents

Prerequisites

  • Snowflake user, preferable of the Service type, with a PAT (Programmatic Access Token) in your Snowflake account. This user must have the authorization to invoke Cortex REST API.

  • Network rule and policy assigned to this user, which allow-lists the IP addresses from where the Mendix application will invoke the API at runtime.

    For applications running locally in Studio Pro, you can retrieve your own IP address from whatismyipaddress.com. For applications running in Mendix Cloud, see Mendix IP Addresses: Mendix Cloud.

  • OpenAI Connector in your Mendix application

Configuring the Snowflake LLM in the OpenAI Connector

Because Snowflake conforms to the API specification of OpenAI, you can use the OpenAI connector to configure the connection to the Snowflake Cortex API.

  1. In OpenAI Connector, create a new configuration by performing the following steps:

    1. Click New.
    2. Enter an identifying name.
    3. For API type, select OpenAI.
    4. For Endpoint, enter your Snowflake account URL with the /api/v2/cortex/v1/ suffix.

    For example, the URL may look like the following: https:/<snowflake_accountname>.snowflakecomputing.com/api/v2/cortex/v1/.

    1. Save your changes.

  2. Add a Large Language Model by performing the following steps:

    1. Hover over the three dots icon and select Manage deployed models from the pop-up menu.
    1. Click Add new model to add an LLM from the list of available LLMs in your Snowflake account.

    Alternatively, you can log into your Snowflake account with Snowsight and select AI & ML > AI Studio > Cortex Playground. The list in the top middle of the screen lists all available models.

    The Model name field in the Deployed Model configuration in Mendix must match exactly with the model name in Snowflake.

    1. Complete the configuration for your model.
    2. Save your changes.

  3. Test the configuration by performing the following steps:

    1. Hover over the three dots icon and select Test from the pop-up menu.
    1. Select your deployed model from the drop-down list.
    2. Click Test.