Graded Prompts

On Graded Prompts, the Graded Badge is the product. Before any prompt reaches the storefront, it goes through a five-step review carried out by a person, not a script.

Step one is a consistency check: the title, description, category, target model, and tags all have to match what the prompt actually does.

Step two examines the instructions — complete, well-structured, every variable clearly labelled.

Step three is a live test: a reviewer runs the prompt on the real model it was written for, trying different parameter values to confirm it holds up across varied inputs.

Step four compares the creator's example outputs to what the model genuinely generates, so previews are never cherry-picked or edited.

Only then does the prompt publish; anything that falls short is returned with specific feedback to fix and resubmit.

What's behind the badge: prompts for 26 AI models across image (Midjourney, Stable Diffusion, FLUX, Ideogram, and more), text (ChatGPT, Claude, Gemini, Llama, DeepSeek, Grok), and video (Sora, Veo, Kling AI, and others). You can browse by category and model, or describe what you want in plain English and let the AI search surface matches.

Free prompts are available to claim; paid ones range from $2.99 to $9.99 and can be bought as a guest without an account. Sellers may also offer a subscription plan — a recurring membership to that one seller's work with a fixed number of prompt claims per month, kept forever once claimed. Each purchase carries a perpetual personal-and-commercial license.

Graded Prompts sells as merchant of record, so if a prompt doesn't perform as described, the refund request goes to the marketplace, not the seller. Creators, meanwhile, keep 85% of each sale.