I built a tool to compare AI models side by side before choosing one

Every time I needed to choose an AI model, I was opening multiple tabs, pasting the same prompt into different tools, and then guessing which one was best. It was slow, inconsistent, and honestly a bit random.

So I built AI AgentLab.

It lets you run one prompt across models like Perplexity, OpenAI, Claude, and Grok at the same time, then compare output quality, speed, token usage, and cost in one place. It also ranks a clear winner so you can make a decision before integrating anything.

You can save your tests and export them as CSV or PDF if you’re working with a team or documenting results.

Right now I’m focused on one thing: figuring out if this actually solves a real problem for other founders and builders.

If you’re building with AI:

• How are you currently choosing which model to use?

• Are you testing across multiple models or just picking one and going with it?

Would really appreciate honest feedback.

Here’s the app:
https://agent-lab.replit.app