AI_SLANG_ENTRY
Model Router Meaning
The cost-control layer that decides which model should answer a request, so a tiny classification job does not always ride first class on the most expensive reasoning model.
What does Model Router mean?
The cost-control layer that decides which model should answer a request, so a tiny classification job does not always ride first class on the most expensive reasoning model.
A model router chooses which model handles a request. Teams use it to balance cost, latency, capability, policy, and fallback behavior across several available models.
Origin and usage
Grew from LLM orchestration patterns and became more concrete as AI platforms shipped model-router features that choose among underlying models by cost, quality, latency, or task fit.
Source type: product-term. Last checked: 2026-07-17.
Current AI platform and architecture term. Microsoft Foundry documents model router as a deployable model that selects an underlying LLM in real time; broader builder usage also covers custom routing layers.
Examples
- The model router sent quick summaries to the cheap model and escalated gnarly planning tasks to the reasoning model.
- A bad model router is just vibes with a billing dashboard.