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

# Testing & Comparison

> Test prompts across multiple models simultaneously and compare results side by side.

Prompt Studio includes built-in tools for testing your prompts against multiple models at once — helping you find the optimal model for your use case before committing to it.

## Single Model Testing

The simplest way to test is clicking **Run** in the editor toolbar. This sends your prompt to the currently selected model and shows the output in the right panel.

## Multi‑Model Testing

For more rigorous testing, use the **Multi-Model Test** feature:

<Steps>
  <Step title="Open Testing">
    Click the **Test** icon in the Prompt Studio sidebar.
  </Step>

  <Step title="Select Models">
    Choose two or more models to test against (e.g., GPT-4o, Claude 3.5 Sonnet, Gemini Pro).
  </Step>

  <Step title="Run">
    Click **Run All** to dispatch the same prompt to all selected models simultaneously.
  </Step>

  <Step title="Compare">
    Review outputs side-by-side with performance metrics for each model.
  </Step>
</Steps>

## Comparison Metrics

For each model response, Traceport displays:

| Metric      | Description                                    |
| ----------- | ---------------------------------------------- |
| **Output**  | The full response content from each model      |
| **Latency** | Response time in milliseconds                  |
| **Tokens**  | Input and output token counts                  |
| **Cost**    | Estimated cost based on the provider's pricing |

<Info>
  Multi-model comparison is also available in the [Playground](/ai-tools/playground), which offers a dedicated side‑by‑side comparison interface.
</Info>

## Batch Testing with Datasets

For systematic testing across many inputs, use **Datasets**:

<Steps>
  <Step title="Create a Dataset">
    Open the **Datasets** section in the Prompt Studio sidebar. Add test cases with variable values and expected outputs.
  </Step>

  <Step title="Run Batch">
    Execute the prompt across all test cases in the dataset.
  </Step>

  <Step title="Review Results">
    Review each test case's input, expected output, and actual output in a table view.
  </Step>
</Steps>

<Tip>
  Datasets are especially useful for regression testing — run them after every prompt change to ensure you haven't introduced quality regressions.
</Tip>