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:
Open Testing
Click the Test icon in the Prompt Studio sidebar.
Select Models
Choose two or more models to test against (e.g., GPT-4o, Claude 3.5 Sonnet, Gemini Pro).
Run
Click Run All to dispatch the same prompt to all selected models simultaneously.
Compare
Review outputs side-by-side with performance metrics for each model.
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 |
Multi-model comparison is also available in the Playground, which offers a dedicated side‑by‑side comparison interface. Batch Testing with Datasets
For systematic testing across many inputs, use Datasets:
Create a Dataset
Open the Datasets section in the Prompt Studio sidebar. Add test cases with variable values and expected outputs.
Run Batch
Execute the prompt across all test cases in the dataset.
Review Results
Review each test case’s input, expected output, and actual output in a table view.
Datasets are especially useful for regression testing — run them after every prompt change to ensure you haven’t introduced quality regressions.
