The AI Playground runs your prompt text against one or more models and shows responses side by side—useful for quality checks before production use.
Routes:
- Personal:
/dashboard/playground - Organization:
/dashboard/organization/{orgId}/playground
You can open the playground from a library prompt or after a Security scanner run (handoff prefill). Compare mode and evaluation panels help spot regressions across models.
Step 1
Enter or load prompt text
Paste instructions or open Playground from a saved prompt. Scanner handoffs may prefill summary and risk context.
Step 2
Select models
Pick from models your plan and **model allowlist** policy permit. Disallowed models return policy errors.
Step 3
Run and compare
Execute runs, inspect latency and token usage where shown, and compare outputs in single or compare view.
Step 4
Iterate
Refine text in place or send findings to **Refine Agent** if the response needs structural improvement.