Smarter-cheaper model selection
Last updated 2026-06-14
Definition
Smarter-cheaper model selection routes each chat turn to a model that fits the job. A simple lookup does not need the heaviest, priciest model, while a hard reasoning question does. Quri picks per turn, so you get a capable answer without paying top rates on every question, and the cost stays proportional to the work.
How to do this in Quri
- Ask questions of any difficulty in /app/chat.
- Let Quri size up each turn and route it to a fitting model.
- Notice quick lookups answer fast, while complex questions get more capable reasoning.
- Check cost per conversation to see how the routing keeps spend proportional.
Frequently asked
- How does Quri choose a model per turn?
- It weighs how demanding the turn is and routes simpler questions to lighter, cheaper models and harder ones to more capable models, so each answer fits its need.
- Does cheaper routing hurt answer quality?
- The aim is to match the model to the task, not to cut corners. Hard questions still get a capable model; only easy ones use a lighter one.