Databricks Unity AI Gateway adds spend controls and budget alerts
Databricks' Unity AI Gateway now offers AI spend controls, allowing users to set budgets and hard caps at various granularities including user, workspace, and organization levels. This feature addresses the challenge of managing unpredictable AI workload costs by providing proactive budget alerts and automated spending limits. The controls are integrated with Unity Catalog and Databricks budgets, offering unified governance for AI usage, cost visibility, and operational accountability across all models and providers, and are available starting today.
- →Configure AI budgets at user, workspace, and organization levels
- →Enforce hard spend caps to prevent runaway AI costs
- →Detailed AI cost analytics via Unity Catalog
- →Unified governance for AI usage and cost management
Features (2) ›
- Configure AI budgets at user, workspace, and organization levels
Unity AI Gateway now allows organizations to set AI spend controls and budget alerts for individual users, specific workspaces, or the entire organization. This feature aims to proactively manage and contain AI costs by establishing predefined limits and notifications.
- Enforce hard spend caps to prevent runaway AI costs
Beyond alerts, Unity AI Gateway offers the ability to enforce hard spend caps. Once a budget threshold is exceeded, the system automatically halts further AI requests until the limit is raised or the next billing period commences, preventing unexpected overspending.
Enhancements (2) ›
- Detailed AI cost analytics via Unity Catalog
The release enhances cost visibility by logging every AI request to Unity Catalog system tables, including DBU costs and token counts. This enables detailed analysis by identity, workspace, model, and provider, offering granular insights into AI spend drivers.
- Unified governance for AI usage and cost management
Unity AI Gateway's spend controls extend Unity Catalog's governance capabilities, providing a unified system for managing AI workloads. This integration allows organizations to govern what agents can do, for whom, and at what cost, ensuring operational accountability.
https://www.databricks.com/blog/introducing-ai-spend-controls-unity-ai-gateway
Related releases
- Databricks Terraform Provider v1.123.0: Improved Plan Validation, Delta Sharing, MLflow, and Bug Fixes Terraform Databricks Provider Releases ·
- Databricks SDK Java v0.137.0 adds AI Gateway and deployment fields Databricks Java SDK Releases ·
- Databricks SDK for Go v0.166.0 Enhances Workspace Services Databricks Go SDK Releases ·
- NorthStar Anesthesia builds clinician scheduling app on Databricks Apps in weeks Databricks Blog ·
- Databricks AI Search adds high-QPS scaling Databricks Blog ·
- AI Applications and Best Practices in Healthcare Databricks Blog ·