hAIgopala › AI terms › Rapid Fire (3 terms per short) › Day 43

Model Routing

MLOps, LLMOps & DeploymentChapter 12: Going LiveTerm 212 of 253

Sending each request to the cheapest model that can handle it.

At Sunrise School

Gopal asks

Can we cut costs further?

Bit answers

Model routing: simple FAQs go to a small, cheap model; complex admission cases go to the big model.

In the real world

Simple FAQ → small model; complex RCA → large model.

Watch the short

Short coming soonSeason 5 · Day 43Follow @haigopala

Quick check

Quick check. Should nightly fee reminders run in real time? Yes or no?

Taught together with

Caching (Semantic Cache) · Batch vs Real-time Inference

Practise it

Try this term and all 253 others in the free Model Training Arena.

Open the Arena →

Follow Bit daily: YouTube English · తెలుగు · हिंदी  |  Instagram English · తెలుగు · हिंदी