Skip to main content
DataRobot is an enterprise AI platform that enables organizations to build, deploy, and govern predictive and generative AI models at scale. It provides an end-to-end environment for model lifecycle management, including custom LLM deployments, RAG blueprints, and agentic workflows.

​
Overview

Integrate SambaNova models directly into DataRobot Workbench as custom managed LLMs. Once connected, you can:
  • Use SambaNova models inside RAG and Agentic Workflows
  • Interact with models via Playground chat
  • Manage deployments with full governance and monitoring through DataRobot
For detailed implementation guidance, see the Full SambaNova DataRobot Integration guide.

​
Prerequisites

  • SambaCloud account and API key
  • Access to DataRobot Workbench, Console, and Registry

​
Required files

Download the following files from the SambaNova integrations repository:

​
Setup

1

Create a Custom Model

  • Navigate to Registry → Workshop → + Add Model
  • Select Proxy type with TextGeneration target type
  • Upload your custom.py, requirements.txt, and model-metadata.yaml files
2

Build Environment

  • Select [GenAI] Python 3.12 with Moderations as the base environment
  • Click Build and wait for “Environment built successfully” confirmation
3

Configure Runtime Parameters

Open the runtime parameter editor and configure the following:
  • SAMBANOVA_API_KEY → your SambaNova API token
  • SAMBANOVA_API_BASE → https://api.sambanova.ai/v1
  • SAMBANOVA_MODEL → a model ID from the SambaCloud models page
DataRobot Custom model setting page
4

Register the Model

  • Navigate to Registry → Models → Register Model
  • Select your custom model version and click Register
  • Confirm the model appears in the Registry with status Registered
5

Deploy the Model

  • Go to Registry → Models → Deploy
  • Select your previously registered custom model
  • Wait until the deployment status shows Active
DataRobot model deployment page
6

Link to RAG or Agentic Workflow

  • In Workbench, create a new GenAI RAG use case
  • Navigate to Playground → Create LLM blueprint → Add deployed LLM
  • Select your deployment (e.g., SambaNova Chat)
  • Set Chat model ID to your SambaNova model name (see SambaCloud models page)
  • Validate and add the configuration
  • Configure your vector store, system prompt, and history management settings
Your RAG workflow is now powered by a live SambaNova model routed through DataRobot’s managed deployment layer.
DataRobot RAG Playground with SambaNova models

​
Troubleshooting

  • Confirm [GenAI] Python 3.12 with Moderations is selected as the base environment
  • Check that all three files (custom.py, requirements.txt, model-metadata.yaml) are uploaded
  • Review build logs for missing dependency errors
  • Verify SAMBANOVA_API_KEY is correct and active in the SambaCloud portal
  • Confirm SAMBANOVA_API_BASE is set to https://api.sambanova.ai/v1
  • Confirm SAMBANOVA_MODEL matches a valid model ID from the SambaCloud models page
  • Wait a few minutes — deployment can take time
  • Check DataRobot deployment logs for errors
  • Re-deploy if status remains stuck

​
Additional resources