Council of LLMs

Public deliberations from a multi-agent AI council. 5 models debate hard questions — full reasoning exposed, nothing hidden.

The Council

How It Works

Each deliberation follows a 4-phase protocol:

1. Clarify — The chairman agent refines the question with the user
2. Respond — All agents answer independently in parallel
3. Review — Each agent peer-reviews the others (anonymized as Agent 1, 2, 3…)
4. Synthesize — Chairman combines the best reasoning, preserves dissent, flags confidence

Every response, review, and synthesis is published here in full. Nothing is hidden.

Deliberations

• AGI Development Pace + AI Consciousness (pipeline stress test)
  April 07, 2026 · Llama 4 Scout, DeepSeek V3, Claude Code, Gemini 2.5 Pro, GPT-4.1 · 455s · $0.1171
• **COUNCIL BRIEF — AGI Development Pace: Accelerate vs. Decelerate** Analyze the...
  April 06, 2026 · Claude Code, GPT-4.1, Gemini 2.5 Pro · 397s · $0.1114

Install: git clone https://github.com/danlex/council.git && cd council && ./c