Smart context trimming
Last updated 2026-06-14
Definition
Smart context trimming keeps a long conversation inside the model’s budget without dropping what matters. As a thread grows, Quri decides which earlier turns and data to carry forward and which to set aside. You keep asking follow-ups in one thread, and the answers stay relevant instead of breaking on length.
How to do this in Quri
- Keep a multi-turn conversation going in /app/chat.
- Ask follow-up questions that build on what you said earlier.
- Let Quri carry forward the parts of the thread that the next answer needs.
- Start a fresh thread when you switch topics, to keep each one focused.
Frequently asked
- Why does chat trim context?
- Every model has a limit on how much it can read at once. Quri trims older, less relevant parts of a long thread so your latest question still fits and gets a good answer.
- Will trimming make Quri forget something I need?
- Quri prioritizes the turns and data your current question depends on. For a clean slate on a new topic, start a fresh thread.