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

# Compare

> A/B test models, temperature, and instructions side by side

**Compare** runs the same prompt against up to three versions of your agent at once, so you can see how different models, temperatures, or instructions respond before you commit to a setup.

Open it from the **Compare** button on the [Overview](/dashboard/overview) page, or from your agent's sidebar.

<img src="https://mintcdn.com/totebot/IEMsjbsyNpqClrGS/assets/dashboard/totebot-ai-compare.png?fit=max&auto=format&n=IEMsjbsyNpqClrGS&q=85&s=d5361722537b285ae904e0cb89f8cf8a" alt="Compare instances side by side" width="2880" height="1800" data-path="assets/dashboard/totebot-ai-compare.png" />

## How it works

* You start with two instances side by side and can **Add an instance** for a third.
* With **Sync** on (the default), one input box sends the same message to every instance at once, so you can compare answers head to head.
* Turn **Sync** off to message each instance independently.
* **Clear all chats** resets the conversations; **Reset** restores the instances to your agent's current settings.

## Tune each instance

Click the gear on any instance to change, just for that column:

* **Model**: pick any model available on your plan.
* **Temperature**: from focused (low) to creative (high). Some models use a fixed temperature.
* **Instructions**: try a different system prompt.
* **Skills**: toggle individual skills on or off.

Changing a setting clears that instance's chat so the next prompt reflects the new configuration.

## What it's for

Use Compare to choose the model with the best balance of quality and cost, to tune tone by testing stricter versus looser instructions, and to confirm a change improves answers before you apply it to your live agent in **Settings → AI**.
