Learn LLMRouter

This section explains how LLMRouter is put together and how to reason about routers, data, training, and deployment.

Note

This docs site is built from the website branch. The runnable code, example configs, and example data live on the main branch.

What you will learn

• How the CLI maps to the core Python modules
• How routers load configs and data
• How router families differ and when to use them
• How training and inference are wired
• How plugins extend the system
• How to deploy and operate routing workflows

Reading map

• Architecture: execution paths, core classes, and data flow
• Router families: categories and trade-offs
• Data and metrics: expected inputs and objective weights
• Training and evaluation: workflows, artifacts, and pitfalls
• Plugin system: custom routers without forking
• Deployment: inference, batch, and integration patterns

Where to go next

• If you want hands-on steps: Quickstart and Tutorials
• If you want a complete list of routers and links: API Reference - Routers
• If you want command flags: API Reference - CLI