Databricks Unity AI Gateway
Databricks governance layer (Mosaic AI / Unity Gateway) managing access, routing, rate limits, logging and guardrails for model serving endpoints via Unity Catalog.
Description
Databricks Unity AI Gateway (formerly Mosaic AI Gateway, now evolving as Unity Gateway) is an enterprise AI governance layer built on Unity Catalog. It centralizes access management, traffic routing, rate limits and cost monitoring for model serving endpoints on the Databricks platform.
Core features include: access and permission control (grant/revoke on Unity Catalog securable objects, foundation model restrictions per group), rate limiting (QPM and TPM limits at endpoint, user and group level), traffic splitting and fallbacks between models on 429/5XX errors, guardrails (PII detection with masking/blocking, unsafe content filters), and observability (request/response payload logging to Delta tables, system tables tracking token usage, latency and cost attribution).
The gateway applies to model serving endpoint types: external models (e.g. OpenAI, Anthropic via bring-your-own-key), Foundation Model APIs (pay-per-token), provisioned throughput and custom models. It also extends governance to AI agents, MCP servers and tools.
Data & KnowledgeData & Knowledge ManagementData connectors, vector database integration, native vector search and data management (PII, provenance, synthetic data).
SecurityEnterprise SecurityCertifications, access controls and data-protection features essential for corporate deployments and cloud privacy compliance.
Developer EcosystemDeveloper EcosystemDeveloper resources: available SDKs, supported programming languages, and infrastructure features and model-deployment methods.
Pricing & Business ModelPricing & Business ModelBilling models (usage-based, provisioned throughput), resource limits and SLA parameters (uptime, support tiers).
Pricing models