Referee.chat is a research tool that has several AI models work on your question while a referee checks the results. You can try it for free, since you get enough credit to cover a match. After that, you can top up any amount you like. It suits questions where a single AI answer isn't enough. 

Multiple models, one verdict  

Think of Referee.chat as hiring a small team of researchers instead of asking one person. In the Panel section, you pick at least two AI models to compete, but you can add up to six. They investigate and discuss the question, then a referee model reviews their work and judges whether the answer holds up. 

You also decide how strict the check should be. Options include publication standard, decision-grade, and formal proof, or you can describe your own bar. The main drawback is cost control. Each round spends credit based on the models you seat, and prices differ, so credits are easy to burn through and hard to budget.

Best for high-stakes questions  

Overall, Referee.chat is a good fit for research, important decisions, or any claim you want double-checked before relying on it. It's less necessary for casual questions, where one AI answer is usually enough. Keep an eye on credits, though, especially if you seat several pricier models.