3Drake

What 3Drake is

3Drake is a completely free tool. You enter a domain or a URL — that is the entire input, and there is nothing else to do. In return you get a written review and a score for that website from several AI models at once.

The core idea: the AI decides everything

There are no guidelines, no checklist and no set of criteria imposed on the model. You hand over the website; everything after that belongs to the model.

The score has no range. It can be 0. It can be 1. It can be 1,000,000. That is the LLM's decision and neither you nor we get to intervene.

The point is to make the model genuinely reason and judge for itself instead of filling in a template. Scoring against a fixed rubric is something any tool can do — what cannot be copied is a model's free judgment.

Multiple models is what makes it special

Every submission is run through Claude, GPT and Gemini at the same time, and more models can be added later. The value is that one input gives you several independent perspectives, instead of you opening each chat yourself and asking one by one.

There is far more to look at than the number. AI can be wrong — so the reasoning, and the way each model chooses to see your website, is the real content here. The score is only the headline.

The leaderboard

Websites are ranked by score, and by its nature the whole thing is probabilistic and random. Anyone has a shot at the top — a small site can land above a famous one, because nothing about the judgment is fixed in advance.

A website that scores highly gets onto the board, and a website on the board can pick up real clicks. That is the reward, and it is also the reason people keep coming back to submit another URL.

How the data works

• Every URL or domain is unique, enforced at the database layer — one website, one entry on the board.
• A score stays fixed for 24 hours. After that the website can be scored again, and the new run replaces the old one.
• There is no ownership verification. If you do not like a result, the way to change it is to have it re-scored after 24 hours.

Transparency is part of the product

This is stated up front rather than left for you to discover: every result here is the subjective opinion of an AI model. A website that is genuinely good can still be given a 0. A rough website can be given a huge number. That is not a bug in the tool — it is the whole experiment.

AI can make mistakes. Read the reviews as opinions, not as an audit.

The deeper layer: this is an experiment on the models

The real subject being tested here is not the websites — it is the models themselves. Websites are just the input material, and that material comes from whoever decides to paste a URL in.

Supplying no criteria is not laziness; it is the condition of the experiment. The moment you give a model a rubric, you are only measuring your own rubric. Remove it, and something else becomes visible: how each model reasons on its own, and what it decides is worth valuing when nobody is guiding it.