Corpus Agentis
The field book to agent ecosystems
The field book to agent ecosystems
Design · Orchestration

LLM Routing (RouteLLM)

A learned router scores each query and sends easy ones to a cheap model, hard ones to a strong model. Trained on human-preference signals to hold quality while cutting spend.

Verified against arxiv.org

How it works

A learned router scores each query and sends easy ones to a cheap model, hard ones to a strong model. Trained on human-preference signals to hold quality while cutting spend.

At a glance
TechniqueRouting
Cost leverMuch lower, up to 85% cost cut at GPT-4-level quality on MT-Bench
Latency leverLower on average, most queries hit the small model
Quality leverNear-frontier when routed well
ExampleLMSYS RouteLLM · Martian · OpenRouter auto
When to use

High-volume mixed-difficulty traffic where most queries are easy and a minority need a frontier model.

Basis
Verified against
https://arxiv.org/abs/2406.18665

The corpus does not rewrite vendor documentation. Where copy is quoted verbatim it is marked as such and attributed to its source.