Introducing Compare: one prompt, every perspective
Send a single prompt to up to four models at once and read the answers side by side. Rolling out to beta members today.
The fastest way to learn what a model is good at is to watch it work next to another one. So we built a mode for exactly that.
Compare sends one prompt to up to four models in parallel and lays the answers out side by side, streaming in real time. You see who answered fastest, who went deepest and who quietly misunderstood the question.
How it works
- Type your prompt as usual.
- Hit
Compareor press⌘ ⇧ Enter, then pick up to four models — or let TunnelGPT suggest a balanced lineup. - Read, vote for the best answer and continue the conversation with the winner. The others stay one click away.
Merge the best of each
Sometimes no single answer is perfect, but together they cover everything. Merge combines the strongest parts of every response into one de-duplicated answer, with a short note on where the models disagreed.
Votes that make Auto smarter
Every vote you cast stays private to your workspace, and it teaches Auto which models you trust for which kinds of task. Compare a few times a week and Auto starts making the same calls you would.
Availability
Compare is rolling out to beta members today and will be part of every plan: Free accounts get five Compare runs a day, Pro and Team get unlimited.
Every model in one place.
Reserve your workspace in early access.