How the room works

One question goes to five frontier models from five different companies. They answer alone, then read each other and argue for three more rounds, and a synthesizer builds the strongest combined answer. Every arc below is one API call, made from this browser tab.

A five-layer diagram. At the top, one question fans out to five model nodes —
Claude, GPT, Gemini, Grok and Muse — with no connections between them, marked
round one is blind. Between each later round every node has a straight line
carrying its own memory forward and four curved lines bringing the other four
models' answers, never its own. A dashed rail down the right collects every turn into
the transcript, which flows into a single synthesis node at the bottom, and one
answer leaves it. — Colour is the model that spoke. Thick straight lines are a model carrying its own memory forward; thin curves are the other models' answers arriving. Download this diagram.

Round one is blind.: No model is told that anyone else is answering, or who. Not merely kept from the others' text — unaware there is a room at all, so nobody postures, defers to whoever they assume is stronger on the topic, or leaves a gap for someone else to fill. The room is introduced in round two, together with the other answers.
Each model keeps its own conversation.: A seat's memory is its own turns plus what it was shown. It never sees its own answer played back as if someone else had said it, and it never sees a round it missed.
An answer cut off at the token ceiling is a failure, not a short answer.: It is marked failed, it does not enter that model's memory, and it is never passed to the others as though it were a position. Half a thought is worse than none, because the room would argue with it.
If a model drops out, the room carries on.: A rate limit or an outage costs you that seat for that round, not the run — the room says who was missing and carries on with the rest. If fewer than two models are still answering there is nothing left to argue with, so it goes straight to the synthesis.
The synthesizer reads the record, not its own side.: It gets fresh context and the whole transcript, and is told not to take a majority vote or average anyone into blandness. It is deliberately not the model that wrote this page.

With all five seated, a four-round run is twenty-one API calls and usually costs a few dollars, billed to you by OpenRouter. Quick mode is one round plus the synthesis, for about a fifth of that. Unseat anyone you do not want in the room.

What should we think about?

The strongest answer the room can support

Keep this conversation

How the room works

Your OpenRouter key

Where your data goes