Sample Outputs

Sample outputs let potential users preview what your prompt can do before they spend credits or time running it.

Strong samples are one of the fastest ways to build trust and increase usage.

What sample outputs are

A sample output is simply:

• A set of inputs you used for a run, and
• The resulting output from the current version of your prompt,

that you've chosen to save and display on your prompt's page.

Users can scroll through these examples to see:

• The style and level of detail your prompt produces.
• How different inputs affect the output.
• Whether the prompt fits their use case.

Creating good samples

To create effective samples:

1. Use realistic inputs
   • Pretend you're one of your target users.
   • Fill in the inputs with concrete, believable data (not just "Test" or "Example").
2. Run the prompt a few times
   • Try a couple of different input combinations.
   • Look for outputs that are both impressive and representative.
3. Save the best runs as samples
   • Pick outputs that clearly show the strength of your prompt.
   • Avoid edge cases or overly niche examples unless that's your main audience.

Remember: users will often decide whether to run your prompt based purely on these samples.

Samples and versions

Samples are always tied to the current version of your prompt:

• If you update the hidden logic, inputs, or model, your existing samples may no longer match behavior.
• After a significant change, you should re‑run and update your samples.

This keeps expectations aligned so users aren't surprised when they run your latest version.

How many samples to include

You don't need dozens of examples—just enough to showcase range and quality:

• 1–2 strong samples can be enough for simple prompts.
• 3–5 is helpful if your prompt supports significantly different use cases or modes.

Focus on clarity and relevance over sheer quantity.

Treat sample outputs as a portfolio of your prompt's best work and an ongoing test suite for changes you make.