Deploy and manage coding agents at scale with the Unity Gateway CLI

The rapid release cycle of frontier models—such as GPT-6, Claude Opus 5.5, and Grok 4.7—has made it difficult for teams to standardize on a single provider without missing better or cheaper alternatives. Supporting multiple agents and providers scatters governance controls, complicates budget tracking, and fragments usage logs. The Unity Gateway CLI aims to resolve this by providing a unified interface for both developers and administrators.

Administrators can centrally configure default models, tools, spending policies, and Smart Routing settings in Unity Gateway. Developers then launch approved agents using simple commands like ug claude or ug codex, with Unity Gateway handling authentication and applying published policies automatically. This allows teams to switch between agents without separate configurations while maintaining consistent governance.

Unity Gateway supports cohort-based rollouts of new models and centrally managed MCP tools and skills, ensuring developers access approved workflows without manual setup. Smart Routing enables cost optimization by assigning simpler tasks to less expensive models and reserving capable models for complex work, with a published evaluation showing 35% cost savings on internal benchmarks.

The CLI also provides spending visibility through ug usage and enables centralized tracing of agent activity to identify inefficiencies like tool failures or oversized responses. Databricks reports using Unity Gateway tracing to fix seven MCP tool bugs, saving an estimated $1.2 million annually in wasted AI spend and lost productivity across its engineering organization.