Consensus Mode

1. Your question goes to every selected model simultaneously — each generates an independent answer without seeing the others.

2. Theo waits for the whole council — each model has its own timeout, and any that fail or time out are recorded and shown per-model. A misbehaving model never blocks the rest.

3. Theo synthesizes and streams the answer— it cherry-picks the strongest reasoning, resolves contradictions, flags uncertainty when models disagree, and streams the result live as it's written.

Below the synthesized answer, a breakdown panel shows:

• Per-model contributions — what each model brought to the final answer.
• Agreement score — 0–100% showing how much the models agree.
• Reasoning narrative — a brief explanation of how the synthesis was done.
• Points of agreement and disagreement — highlighted for transparency.

Expand the Council Viewto see each model's full, unedited response alongside its latency and token count. This lets you compare how different models approached your question.