kvsdileep

kvsdileep / model-council

Public

One prompt. Four frontier models. A structured debate. One synthesized verdict.

23
25
69% credibility
Found Apr 13, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

Model Council is a web app where users input a prompt to watch four leading AI models debate in structured rounds before receiving a synthesized final verdict.

How It Works

1
🔍 Discover Model Council

You find this fun web app that lets top AI thinkers debate your questions like a council meeting.

2
💻 Start the App

You get the app running on your computer and open it in your web browser to see the welcoming screen.

3
💭 Ask Your Question

You type a question or topic into the big text box, like 'What's the best way to learn guitar?', and choose options like adding web info.

4
🚀 Convene the Council

You click the button to gather four smart AIs, and they start sharing their independent thoughts in the first round.

5
🗣️ Watch the Debate

You see the AIs critique each other, update their views, and maybe go to a final round if they still disagree, with words appearing live.

6
⚖️ Get the Verdict

A judge AI reviews everything and shows a table of agreements, disagreements, confidences, plus a clear final answer.

📥 Save Your Results

You download a neat summary file to keep the full debate and verdict for later.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 23 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is model-council?

Model Council is a TypeScript Next.js app that runs your prompt through four frontier models—Claude Sonnet 4.6, GPT-5, Gemini 3.1 Pro, and Grok 4.20—in a live, structured debate. Enter one prompt via the web UI, toggle web search or force a third round, hit convene, and stream responses: independent answers, mutual critiques, optional finals, then a Claude Opus verdict with agree/disagree tables and synthesized answer. It tackles flaky AI outputs by making models argue, converging on stronger verdicts or exposing rifts—export sessions to Markdown for records.

Why is it gaining traction?

Live token streaming in slick model cards turns dry prompting into a spectator sport, with auto Round 3 on disagreements adding tension absent in basic parallel callers. One-click GitHub deploys via Vercel hook devs needing quick model council AI debates without API wrangling or custom orchestration. Markdown exports fuel one prompt a day journals or model council GitHub shares.

Who should use this?

Prompt engineers benchmarking LLMs on ambiguous queries, AI researchers dissecting model biases in real-time debates, or product leads seeking explainable consensus for decisions like model council perplexity evals. Suits solo tinkerers over teams building from scratch.

Verdict

At 19 stars and 0.7% credibility, this v0.1 skeleton lacks polish and tests but deploys fast with an OpenRouter key—grab it for model council GitHub experiments if debate formats click. Fork and harden for anything beyond prototypes.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.