NewYour first question is free, no login

Not a chatbot. A council.

Ask a hard question and three AI models argue it: one for, one against, one on how to make it better. A fourth weighs them into one verdict, with the disagreement left in view.

3 + 1

Three models argue, a fourth decides

~30s

Typical wait for a full verdict

Runs on models from

OpenAI GPTAnthropic ClaudeGoogle GeminiMeta LlamaAlibaba QwenOpenAI GPTAnthropic ClaudeGoogle GeminiMeta LlamaAlibaba Qwen

Start hereTake two minutes to read the site before you ask. Once you know how the panel argues and how the chairman decides, you will ask sharper questions and get far more useful verdicts.

// Three roles

Every question, three sides.

For itStrongest caseBest version

The Advocate

The case for it. Argues the strongest honest version of yes, so the best case gets made properly rather than assumed.

Against itFailure modesBlind spots

The Critic

The case against it. Finds what breaks before you do. This is the seat most single-model answers are missing.

Next stepImprovementsAction

The Builder

How to make it better. Turns the argument into a next step instead of leaving you with two opposed opinions.

// How it works

From question to verdict.

Four steps, in the same order, every time. The panel is picked for the kind of question you ask.

01

The gate

Council decides whether your question has one correct answer. Factual ones get a single line and an honest nudge, not a staged debate.

02

Three roles

Judgment questions convene a panel picked for the type of question: one argues for, one against, one works out how to make it better. They answer in parallel.

03

The chairman

A fourth model reads all three, picks the strongest reasoning and issues one verdict, with the disagreement left in view.

04

Verification

Planned for Plus: factual claims in the verdict get checked against an outside source. Not switched on in the free preview yet.

It argues with itself.

It picks the strongest case.

It commits to one answer.

It shows you the reasoning.

It refuses to fake certainty.

// See it in action

Watch them argue, then decide.

Pick a question and watch the council work it through: three labs, three angles, one verdict.

GPTThe case for

If your mock scores were climbing in the last two months, a focused year can close the gap. A better college shapes your peers, internships and placements for the next decade, so the upside is long-lived.

ClaudeThe case against

A drop year is a full year of pressure with no guarantee. Many repeaters plateau, and the cost is real: coaching fees, a year behind your batch, and your health. If the problem was consistency rather than knowledge, another year may not fix it.

GeminiHow to improve it

Here is what I would do: take the seat you have and keep preparing alongside it. Sit one full mock every fortnight, and if your rank clears your target by December, attempt again with college as your safety net.

The chairman’s verdict Medium confidence

Drop only if your mocks were rising and you can name exactly what you will do differently. Otherwise take the seat and keep your options open.

Where they disagree: Whether a better college is worth a year of risk.

Example debate, written to show the format. Ask your own below and the real models answer live.

Compare any two models

GPT, Claude, Gemini, DeepSeek and more, side by side on reasoning, coding, writing, research, speed and value.

// What we hold to

We debate before we believe.

Council principle

“It knows when to shut up.”

Ask something with one correct answer and you get one line, not a staged debate. You are not billed for theatre.

Council principle

“It refuses to fake certainty.”

The same confident tone for a settled fact and a coin flip is the problem Council exists to fix.

Council principle

“Every question has three sides.”

For it, against it, and how to make it better. The verdict comes with the split left in view.

// Group chat New

A group chat that reads before you sign.

Talk to the whole council at once. When the message moves money, three AIs from different labs agree on exactly what you meant and flag anything risky, so nothing is signed by mistake.

  • One room, three labsThe Advocate, the Critic and the Builder each come from a different AI lab. They read each other’s messages and push back by name.
  • Reads money the way you meant itType a transfer, swap or payment in plain words. The council agrees on the exact amount, asset, network and recipient.
  • Catches what one model missesA vague “half”, a name instead of an address, a missing time or slippage: the Critic flags it before anything is signed.
  • One final readingThe Chair returns the exact steps, a confidence level, and whether it is ready to sign or needs one more answer.

Council never touches your wallet or bank. It only reads the instruction.

Example · crypto transfer

Council group chat

Checking a transaction

Building a fintech or crypto product? See Council for business →

// Pricing

One council. Pick how much you need.

Every plan runs the same mix of models from different labs, and no model is locked behind a higher plan. Right now Council is free with no sign-up, up to 25 questions a day; paid plans open when accounts launch.

Live nowComing soon

Free

Try a real council before you pay for anything.

₹0/mo

Free forever

Ask a question
  • Three models from three labs on every question (live now)
  • One verdict, with the disagreement shown (live now)
  • A model receipt on every answer (live now)
  • Follow-up questions in the same conversation (live now)
  • 3 full councils a month once accounts open (coming soon)
  • Web research for recent events (coming soon)
  • MCP access is part of Ultra

Plus

Bring your own work and let the council pull it apart.

₹1,799/mo

Billed monthly

Get notified
  • Councils of up to 7 models, from every lab (coming soon)
  • Every premium model, none held back (coming soon)
  • One combined verdict, disagreements listed (live now)
  • Independent answers: no model sees another’s (live now)
  • Blind review: models rate each other anonymously (coming soon)
  • Export as Word, PDF or Markdown (coming soon)
  • Attach files: 4 a message, 10 MB, 500 a month (coming soon)
  • Bring your memory from ChatGPT or Claude (coming soon)
  • Saved councils, 3 projects, custom instructions (coming soon)
Most popular

Pro

For people who bring the council work every day.

₹2,999/mo

Billed monthly

Get notified
  • Everything in Plus, and:
  • Councils of up to 10 models (coming soon)
  • Twice the monthly budget of Plus (coming soon)
  • 1,000 uploads a month, 5 a message, 20 MB (coming soon)
  • Verdict read aloud, a different voice per seat (coming soon)
  • 5 projects (coming soon)
  • Priority support (coming soon)
Flagship

Ultra

Your documents inside the council, and the council inside your tools.

₹9,999/mo

Billed monthly

Get notified
  • Everything in Pro, and:
  • Your document library, answered with page-level citations (coming soon)
  • Recording review that cites the exact moment, like 14:32 (coming soon)
  • MCP server for Claude, Cursor and Windsurf (free beta now) (live now)
  • The largest monthly budget (coming soon)
  • 2,000 uploads a month, 10 a message, 50 MB (coming soon)
  • Unlimited projects and a priority queue (coming soon)
  • Early access to new models (coming soon)

Team

One bill and a shared budget for a group.

Custom

Priced after we talk

Talk to us
  • Several people on one bill (coming soon)
  • A larger shared monthly budget (coming soon)
  • Invoices instead of card checkout (coming soon)
  • Terms agreed in writing (coming soon)

Prices are what we plan to charge and may change before billing opens. Compare every feature

// Test your council now

Ask the council anything.

Three angles, weighed into one verdict. Your first question is free, no login.

3 models, about 30 seconds

The council panel

The case forGPTOpenAI, via OpenAI API
The case againstClaudeAnthropic, via Anthropic API
How to improve itGeminiGoogle, via Google AI Studio
ChairmanQwenAlibaba, via Groq

GPT, Claude and Gemini argue; Qwen chairs. If a model is busy or today’s budget for the paid models is used up, a free backup from another lab steps in, and every answer shows exactly which model wrote it.

Want a record of your debates? Sign up.