PromptJury is an LLM evaluation platform that lets users compare prompts and models such as GPT, Claude, and Gemini using realistic test cases and blind jury grading. It generates test cases and rival prompt drafts automatically, then scores every answer using a panel of models from different makers where no model grades its own output. Users set the grading criteria before any prompt is written, ensuring the evaluation standard is fixed and unbiased. Results include per-model scores with confidence ranges, cost per 1,000 calls, and a clear recommendation on which prompt and model to ship.