Skip to main content

Deployment

Sutro supports managed and self-hosted deployments. Provider keys, cloud credentials, and custom tools are configured per deployment.

Deployment credentials

Each deployment has its own URL and API keys. Configure them with environment variables, the CLI, or the Python SDK as described in Authentication. Changing deployment URLs clears a saved SDK key so credentials are not sent to the wrong deployment.

Data retention

Sutro supports configuring custom data retention policies on an as-needed basis. Our current default retention policy for the Batch inference API is 90 days.

Contact

Contact [email protected] for self-hosting, BYOK/BYOC, multimodal extensions, data retention, or custom model and tool requirements.