Consensus Mode

1. Your question goes to all 6 models simultaneously — each generates an independent answer without seeing the others.

2. Quorum resolves when 4 of 6 respond— the system doesn't wait for slowpokes. Models that take too long are recorded as timeouts.

3. A synthesis engine combines the results — it cherry-picks the strongest reasoning, resolves contradictions, and flags uncertainty when models disagree.

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.