> ## Documentation Index
> Fetch the complete documentation index at: https://docs.intelligence-management-platform.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Model override and usage

> Switch the AI model for a conversation and inspect token and cost usage per message.

export const RightsCallout = ({right, flags}) => <Info>
    Who sees this: requires {right === "none" ? "no special rights — available to every signed-in user" : `the ${right} right (or super admin)`}.
    {flags && flags !== "none" ? ` Only visible when the server enables ${flags}.` : ""}
  </Info>;

<RightsCallout right="none" flags="none" />

Each agent has a default model configured by its administrator. You can switch to any other available model for the current conversation from the header menu, and inspect token and cost details in the Usage panel.

## Switch the model

<Steps>
  <Step title="Open the menu">
    Click the **⋯** button in the conversation header (or press <kbd>⌘</kbd>+<kbd>.</kbd>) to open the conversation actions menu.
  </Step>

  <Step title="Choose Model…">
    Click **Model…**. If a model override is already active, the menu entry reads **Model… (`{name}`)** showing the currently selected model.
  </Step>

  <Step title="Pick a model">
    The **Model** dialog lists every active model available on this IMP instance. The agent's configured model is marked **(default)**. If the list is long, a search box appears — type to filter.
  </Step>

  <Step title="Confirm">
    Click any model to select it. The dialog closes and a **Model: `{name}`** badge appears in the conversation header to confirm the override is active. A note in the dialog reads: "Applies to new messages in this conversation only."
  </Step>
</Steps>

To return to the agent's default model, open the dialog again and select the **(default)** model. The badge disappears from the header when no override is in effect.

<Note>
  Model overrides apply only to the current conversation. Starting a new chat always uses the agent's configured default model.
</Note>

## View token and cost usage

Click the context-usage chip in the header — for example **9% · 12k** — or open **Usage** from the **⋯** menu. The **Usage** panel opens with the title "Usage" and the description "Token and budget usage for this conversation."

The panel is divided into sections:

**Context window** — a bar and label showing how much of the current model's context window the conversation occupies (for example, "9% of 200k tokens"). This reflects live occupancy across all the messages currently in the model's context.

**Session usage (cumulative)** — token counts accumulated over the full session:

| Row                 | What it counts                                                             |
| ------------------- | -------------------------------------------------------------------------- |
| **Total tokens**    | Sum of input and output tokens across all messages in this conversation.   |
| Input tokens        | Tokens sent to the model (your messages, retrieved context, tool outputs). |
| Output tokens       | Tokens generated by the model.                                             |
| Reasoning tokens    | Tokens consumed by models that expose reasoning steps.                     |
| Cached input tokens | Input tokens served from the provider's prompt cache (reducing cost).      |

A plain-language note in the panel explains: "Tokens are the units AI usage is measured in — roughly ¾ of a word each."

If your account has a budget, a **Your budget** section appears below the token breakdown with a bar showing how much of your period budget has been used.

<Tip>
  The cumulative token count grows with every message. Long conversations with large tool outputs accumulate quickly — see [Context compaction](/user-guide/chat/context-compaction) to keep the conversation within the model's limits.
</Tip>

## Next steps

<Columns cols={2}>
  <Card title="Context compaction" icon="archive" href="/user-guide/chat/context-compaction">
    Compact long conversations to stay within the context window.
  </Card>

  <Card title="Budgets in chat" icon="circle-dollar-sign" href="/user-guide/chat/budgets-in-chat">
    Track your spending budget in the top bar.
  </Card>
</Columns>