Skip to main content
Honeydew can integrate with a number of LLM providers and platforms. Two modes of operation are possible:
  • Using a model platform (e.g Snowflake Cortex, Databricks Mosaic, Amazon Bedrock, etc).
  • Using the model provider directly (e.g. OpenAI, Anthropic, Gemini, etc).
The decision which mode to use depends on your architecture, security and compliance requirements, as well as cost and budget considerations.

​
Anthropic Claude

Anthropic Claude Sonnet 5 is the default LLM model used by Honeydew AI for text-to-data and data analysis.

​
Direct Integration

Honeydew supports direct integration with Anthropic Claude models using the Anthropic API, and with OpenAI GPT models using the OpenAI API. To configure it, contact Honeydew support with your Anthropic or OpenAI API key.

​
Snowflake Cortex

Honeydew supports using Anthropic Claude, OpenAI GPT and Google Gemini models via the Snowflake Cortex REST API. Honeydew calls the models through the Snowflake connection. The role Honeydew runs with needs Cortex REST API access. See Snowflake Cortex Requirements.

​
Databricks Mosaic

Honeydew supports using Anthropic Claude, OpenAI GPT and Google Gemini models via Databricks Foundation Model APIs. Honeydew calls the models through the Databricks connection. The connection’s user or service principal must be able to query the model serving endpoints.

​
Amazon Bedrock

Honeydew supports using Anthropic Claude and OpenAI GPT models via Amazon Bedrock (Beta) in your AWS account. To configure it, contact Honeydew support with a Bedrock API key and its AWS region. Honeydew uses global cross-region inference by default. For data residency, ask Honeydew support to use a geographic inference profile (such as us or eu) instead.