Skip to main content

Databricks

Connect to models hosted on Databricks Model Serving.

Configuration

FieldRequiredExample
Application NameYesmy-databricks-app
Base URLYeshttps://adb-<workspace-id>.<region>.azuredatabricks.net/serving-endpoints
Model NameYesdatabricks-meta-llama-3-1-405b-instruct
API KeyYesYour Databricks token

Getting Your Credentials

  1. Log in to your Databricks workspace
  2. Click your username → User Settings
  3. Go to DeveloperAccess tokens
  4. Generate a new token

Base URL Format

https://adb-<workspace-id>.<region>.azuredatabricks.net/serving-endpoints

Replace:

  • <workspace-id>: Your Databricks workspace ID
  • <region>: Your region (e.g., westus, eastus2)

Models depend on your Databricks workspace configuration. Common options include:

ModelDescription
databricks-meta-llama-3-1-405b-instructLlama 3.1 405B
databricks-meta-llama-3-1-70b-instructLlama 3.1 70B
databricks-dbrx-instructDatabricks DBRX

For more information, visit the Databricks Model Serving Documentation and Foundation Models Overview.

Advanced Parameters

ParameterTypeDescription
max_tokensintegerMaximum tokens
temperaturefloatRandomness
top_pfloatNucleus sampling
top_kintegerTop-k sampling

Setup Steps

  1. Navigate to AI Applications → New Application
  2. Select Model Providers tab
  3. Click Databricks
  4. Enter your Application Name
  5. Enter your Base URL (Databricks serving endpoint)
  6. Enter your Model Name
  7. Enter your API Key (Databricks access token)
  8. Configure Advanced Settings (optional)
  9. Click Test Response to verify (optional)
  10. Review and submit