Ask three models the same question
Let Me AI That puts Claude, Gemini and ChatGPT side by side on one prompt, so you can see where they agree, where they diverge, and which one actually answered what you asked.
Why compare at all
The gap between frontier models is narrower than it has ever been, and that makes the question "which model is best" close to meaningless. What matters is which one is best at your task, in your phrasing, with your constraints. That is not something a benchmark table can tell you, because your work is not a benchmark.
Running one prompt through three models takes about fifteen seconds and answers the question directly. Sometimes all three agree and you can trust the answer more. Sometimes one of them quietly invents a detail the other two do not, which is the most useful thing this tool does.
Bring your own key
This site has no account system, no subscription and no free credits pool, because it never pays for a single token. You supply your own API key for whichever providers you want to use, and the request goes from your browser straight to Anthropic, Google or OpenAI.
What that means in practice
- Your keys stay in your browser. They are stored in local storage on your own device and are never transmitted to this site's server.
- There is no server to leak them from. This site runs as a Cloudflare Worker that serves static files and shortens links. It has no endpoint that accepts an API key, and no database of users.
- You are billed directly by each provider at their normal rates. Nothing is marked up, because nothing passes through here.
- Your prompts are not logged. They go to the provider you chose and nowhere else.
The trade is that you need an API key, which is a small amount of setup. In exchange you get no rate limits beyond your own account's, no queue, and no company in the middle reading your prompts.
Getting keys
- Anthropic Console — for Claude
- Google AI Studio — for Gemini
- OpenAI Platform — for ChatGPT
You do not need all three. The tool enables whichever panels have a key and leaves the rest switched off.
What it does not do
It does not judge the answers for you, score them, or pick a winner. Three answers to one question is the product; the judgement is yours. It also does not store conversations anywhere — your history lives in your browser and the Export button writes a plain text file to your machine.
Who made it
An independent side project, not affiliated with Anthropic, Google or OpenAI. Model names and trademarks belong to their respective owners.