How do I optimize a prompt?
Most weak prompts aren't wrong, they're thin. "Write a blog post about coffee" leaves the model to guess the reader, the length, the angle, and the format, and it guesses the most generic option every time. Optimizing means filling those gaps on purpose.
- Paste your prompt. Any prompt up to 3,000 characters.
- Pick the type. Text (ChatGPT, Claude, Gemini), image, or video. Add a goal if you have one.
- Read the score. A total out of 100, a score per criterion, and the issues that matter most.
- Use the fixed prompt. Fill any [placeholders] and paste it into your AI tool.
What does a high-scoring prompt look like?
Before (scores low): write a blog post about coffee.
After: Write an 800-word blog post for [home baristas who just bought their first espresso machine] explaining how grind size changes taste. Friendly, plain English. Use H2 subheadings, one short example per section, and end with a 3-item checklist. Same intent, but the model no longer has to guess.
Prompt optimizer or prompt rewriter?
Want to know why a prompt isn't working? Use this page. Already know what you want and just need it shaped for Midjourney, Flux, Stable Diffusion, or Veo? Use the AI prompt rewriter. Building a structured prompt from scratch? Try the JSON prompt generator. Writing prompts for ChatGPT? The ChatGPT prompt formulas guide has templates you can reuse.
Prompt optimizer FAQ
What is a prompt optimizer?
A prompt optimizer checks a prompt for what's missing and rewrites it so the AI has less to guess. This one grades your prompt out of 100 on five criteria, lists the problems, and gives you a fixed version. It works for ChatGPT-style text prompts, image prompts, and video prompts.
How is the score calculated?
Five criteria, 20 points each. Text prompts: clear task, context, constraints, output format, examples or audience. Image prompts: subject, setting, style, lighting and color, composition and camera. Video prompts: subject and action, scene, camera move, lighting and mood, pacing and length. An AI model does the grading, so treat the number as a guide, not a measurement.
What's the difference between the prompt optimizer and the prompt rewriter?
The optimizer is a check-up: it scores your prompt and tells you what's weak, then fixes it. The prompt rewriter skips the grading and reshapes a prompt for one specific model, like Midjourney flags or a Stable Diffusion negative prompt. Use the optimizer when you want to learn why a prompt isn't working.
Why does the fixed prompt have [brackets] in it?
Those are details only you know, like your audience, a word count, or a brand name. The tool won't invent them. Fill them in before you paste the prompt.
Is the prompt optimizer free?
Yes. No signup, no card. A pace limit and a daily cap keep it free.