39 lines
1.3 KiB
Plaintext
39 lines
1.3 KiB
Plaintext
|
---
|
||
|
|
title: Context
|
||
|
|
description: See how much of the model's memory a session is using.
|
||
|
|
---
|
||
|
|
|
||
|
|
# Context
|
||
|
|
|
||
|
|
Every model can only hold so much of a conversation at once — its context. OpenChamber shows how full that is, so you can tell when a session is getting close to the limit and a reply might start dropping older detail.
|
||
|
|
|
||
|
|
## The quick indicator
|
||
|
|
|
||
|
|
While you chat, a small gauge shows the percentage of context used. It changes color as you fill up:
|
||
|
|
|
||
|
|
- green — plenty of room
|
||
|
|
- yellow — getting full (around three-quarters)
|
||
|
|
- red — nearly full
|
||
|
|
|
||
|
|
Hover it (or tap it on mobile) to see exact token counts.
|
||
|
|
|
||
|
|
## The full context panel
|
||
|
|
|
||
|
|
Open the **Context** tab in the right sidebar for a fuller picture of the current session:
|
||
|
|
|
||
|
|
- the model in use and when the session started
|
||
|
|
- total tokens against the model's limit
|
||
|
|
- message and cost totals
|
||
|
|
- a breakdown of the last reply's tokens
|
||
|
|
- a rough split of what's taking up context (your messages, the agent's, tool output)
|
||
|
|
|
||
|
|
The breakdown is an estimate, not an exact count — use it to spot what's filling the window, not for billing.
|
||
|
|
|
||
|
|
## What to do when it's full
|
||
|
|
|
||
|
|
Start a fresh session for a new task instead of letting one session grow forever. A shorter context is faster and keeps the model focused.
|
||
|
|
|
||
|
|
## Related
|
||
|
|
|
||
|
|
- [Projects](/projects/) — sessions are grouped per project
|