Litmiq

Litmiq is a browser-based prompt evaluator that compares OpenAI model outputs side by side on your own test cases. Key features: • Run one system prompt across multiple OpenAI models at once • Add custom test cases with optional expected behavior • Score every response with per-metric LLM judge ratings • Edit scoring criteria to match your quality bar • Save setup automatically so you can pick up where you left off • Export complete results as JSON. Litmiq runs entirely in your browser with no signup and no fee for the tool itself. Your OpenAI API key is stored in localStorage and sent only directly to OpenAI's API, so it never passes through a Litmiq server. OpenAI bills your API usage at standard rates. Judging uses GPT-5.4 Mini, which typically costs under one cent per eval run. The results panel shows per-metric scores for each model and test case, so you can see exactly where a model follows instructions and where it falls short. Built for developers, prompt engineers, and product teams who need to choose a model or validate a system prompt before shipping. Use it to check customer support replies, policy answers, and edge cases such as address changes, then export the JSON to share results or feed them into your own reporting.