# Deploy and manage coding agents at scale with the Unity Gateway CLI > Source: > Published: 2026-09-24 16:46:05+00:00 Let your developers use their coding agent of choice while keeping your organization at the cost-performance frontier In the last six months, GPT-6, Claude Opus 5.5, Gemini 3.8, and Grok 4.7 all shipped, alongside strong open-weight models like Kimi K3, GLM-5, and DeepSeek V4.1. There’s a new frontier model roughly every five days, and as the cost-performance frontier keeps moving, the model that was the best choice for a job a few months ago often has a better or cheaper alternative today. That pace puts the teams that own developer tooling in a squeeze. Standardize on one provider, and you risk missing the better models or lower costs available elsewhere. Support many agents and providers, and you give engineers a real choice, but scatter access, budget controls, security policies, and usage logs across separate systems. What organizations actually need is both: the freedom to adopt the best option whenever it appears, and a consistent way to govern it across thousands of developers. **Today, we're introducing the** [**Unity Gateway CLI**](https://github.com/databricks/unity-gateway) to deliver both. In [Unity Gateway](https://www.databricks.com/product/artificial-intelligence/unity-gateway), admins manage approved models, tools, and spending policies from one place. Developers keep working the way they prefer, launching any approved agent with a command like ug claude or ug codex, while Unity Gateway handles authentication and applies the published settings for them. When a better model or a more efficient workflow emerges, teams roll it out across every supported agent at once. The result is a coding setup that keeps pace with the cost-performance frontier as it moves. Unity Gateway now provides admins with a single place to configure coding agents, including their default models, MCP servers, skills, Smart Routing, and spending policies. Admins can manage these settings in **Unity Gateway → Govern → Agent Configuration**. Once admins publish an agent configuration, developers can connect to any supported coding agent with a single command. The [Unity Gateway CLI](https://github.com/databricks/unity-gateway) handles authentication, connects the agent to Unity Gateway, and applies the published settings before opening the agent’s familiar interface: Developers can switch between enabled agents without configuring each one separately. Organizations can deploy ug through device management so everyone starts with the same approved setup, with settings that admins can lock. “With the Unity Gateway CLI, we can route all coding-agent model and tool traffic through a single governed path while maintaining identity-level attribution and access to approved models and MCP tools. Since rolling it out, our coding agents have generated more than 61 billion input tokens across roughly 360,000 requests, with centralized visibility into usage and spend.”  —John Xing, Chief Technology Officer, Concurrence As new agents and models become available, admins can choose the setup they want developers to start with and publish it to the entire team with a single configuration change. The next time developers run ug, the CLI automatically syncs the configuration change and launches the recommended agent with the selected model in its familiar interface. Admins can update that choice as quality, cost, and team needs change. They can also set-up more sophisticated policies such as cohort-based roll outs of new models. Admins can configure shared MCP tools and skills and make them available through supported agents. Developers gain access to the approved tools and proven workflows within their configured agents, without having to set each one up separately. Coding tasks vary in complexity. Admins can enable Smart Routing centrally so simpler work can use less expensive models while harder work gets more capable ones. Developers stay in their preferred coding agents without having to choose the right model for each task. Smart Routing selects a model for the main session and can make a separate choice for work that the agent delegates to a subagent. In our [published Smart Routing evaluation](https://www.databricks.com/blog/smart-routing-unity-ai-gateway-match-frontier-quality-30-lower-cost-task), this approach delivered **35% cost savings on our internal coding benchmark**. Unity Gateway lets teams recommend lower-cost agents and models as spending reaches defined budget thresholds. These become the defaults for new launches without interrupting ongoing work. Developers can check spending and remaining budget with ug usage. Unity Gateway can capture traces from coding agents, including local tool calls and skill invocations. Admins enable tracing centrally, and ug configures supported clients to export those traces to the lakehouse’s [unified trace table](https://docs.databricks.com/aws/en/ai-gateway/unified-trace-table). Teams can then use Genie to find repeated tool failures, oversized responses, and other sources of wasted tokens, fix them, and measure the results. In our own engineering organization, we [used Unity Gateway tracing with Genie One](https://www.databricks.com/blog/how-we-eliminated-1-million-year-wasted-ai-agent-spend-one-hour) to find and fix seven MCP tool bugs, which saved us an estimated $1.2 million per year in wasted AI spend and lost productivity. At Databricks, the Unity Gateway CLI helps us roll out new models and cost optimizations across our engineering organization without changing how developers work. Get started with the [Unity Gateway CLI quickstart](https://docs.databricks.com/aws/en/ai-gateway/coding-agent-quickstart), or dive deeper by [configuring your coding agents in Unity Gateway](https://docs.databricks.com/aws/en/ai-gateway/coding-agent-configure-govern). Subscribe to our blog and get the latest posts delivered to your inbox.