> ## Documentation index
> Fetch the complete documentation index at: https://docs.prescientai.com/llms.txt
> Use this file to discover all available pages before exploring further. Add `.md` to any page URL to get that page as Markdown.

# Compare two configs

> Open Model Comparison, select two model configs, and read the Preference Score, the fit, and the modeled metrics of each one.

Model Comparison shows two configs of your models side by side. Use it to see which config the comparison prefers and how the modeled results of the two configs differ.

In a label on this page, a word in braces is a value from your account. The efficiency is the efficiency metric of the target, for example ROAS.

## Open Model Comparison

Model Comparison has no item in the sidebar. Open it from one of two places.

- On a model detail page in Model Center, select **Compare configs**. The comparison opens in a new browser tab, with the **Default** config as **Model A**.
- On the Attribution page, open the **Model** control and select **Compare All Models**. This item shows only when your workspace has more configs than models.

> **If the page has no comparisons:** The page shows **No comparisons available yet** when your workspace has no configs to compare. The message says that comparisons appear once your Prescient team has built comparable model configurations.

## Select the two configs

1. In the **Comparing Models** card, select a config in **Model A**.

2. Select a config in **Model B**.

   **Model B** lists first the configs that have a comparison with **Model A**. Other configs are under **Models in other comparison sets**. If you select one of them, the page changes **Model A** to a config that it can be compared with.

   If the page shows **No comparison models available for {model}.**, select a different config in **Model A**.

3. In the page header, select the target.

4. In the page header, select the timeframe.

   A timeframe that you cannot select has no comparison for the two configs.

## Read which config is preferred

The **Model Selection Criteria** section shows one panel for each config.

1. Read the **Preference Score** of each config. The score runs from 0 to 100.

2. Find the **Recommended** badge. The badge is on the config that the comparison prefers.

3. Read the result under the panels: **Model A Preferred**, **Model B Preferred**, or **Equally Preferred**, with a short explanation.

4. Read the rows under **Additional Metrics** to see the criteria behind the score. See [Model Comparison reference](https://docs.prescientai.com/reference/model-center/model-comparison#additional-metrics).

> **The comparison uses one model run:** Each panel shows **Run compared {date}**, the date of the model run that the comparison used. If the panel also shows **Latest run {date}**, the config has a later run, and the figures of the comparison describe the earlier run. Model Center shows the latest run.

## Read how the results differ

1. In **MMM Fit**, compare the **Model A** line and the **Model B** line with **Reported Total**. Select a series in the legend to hide or show the series.

2. In **MMM Metric Comparison**, read the value of each config and the **Difference** row. The difference is the change from **Model A** to **Model B**, as a percentage.

3. In **Historical {efficiency}**, read the modeled efficiency of each config by channel. The table has one group of columns for each timeframe.

## Next steps

- [Model Comparison reference](https://docs.prescientai.com/reference/model-center/model-comparison): each section, criterion, and message.
- [Configs and the Preference Score](https://docs.prescientai.com/model-center/configs-and-the-preference-score): what a config is and how to read the score.