Every tier of RefNode runs the same ritual. Your question is put to a council of artificial advisors drawn from rival laboratories. Each answers alone, seeing nothing of the others — so no one anchors, no one follows. Then the answers are laid open, and each advisor confronts the rest: conceding what is stronger, defending what survives, naming what the others missed. On the higher tiers, a fourth mind that held no seat reads the entire exchange and renders a verdict. That is the whole machine. The ladder below decides only two things: who sits at your table, and how often you can convene them.
What happens when you ask
- The Blind Round Every advisor answers your question independently and simultaneously. None can see another's response — independence is the condition everything else depends on.
- Cross-Examination The answers are unsealed. Each advisor reads the others, agrees where agreement is earned, objects where it isn't, and refines its own position under challenge.
- The Verdict On Pro and Max, a fourth model — one that held no seat and argued no side — weighs the full deliberation and renders the Chair Brief: what the council concluded, and how much weight it can bear.
Four rungs, one method
- Free Proves the method. One council of capable open models deliberates your question in full — blind round, cross-examination, the works. No account, no card, no setup. Enough to feel why three independent minds beat one confident one. Up to 20 Free Council asks a day
- Basic · $9.99 / mo Seats the experts. Three specialist councils, each tuned to a different kind of thinking — balanced judgment, global perspective, and deliberate step-by-step reasoning. Pick the council that fits the question. Daily model-call limit · monthly AI credits included
- Pro · $19.99 / mo Renders verdicts. Three synthesized councils, seating models from the leading laboratories. Every deliberation now ends in a Chair Brief — a fourth model's written verdict on what the council concluded and where it fractured. Daily model-call limit · monthly AI credits included
- Max · $29.99 / mo Convenes the war room. The structured Workbench: frontier advisors produce evidence-ledgered, independently structured analyses, cross-examine, and maintain a standing decision memo across long-running project sessions. Built for decisions that take more than one sitting. Runs on your AI credits. Highest daily model-call allowance · Workbench on your AI credits
- Recharge Keeps the council in session. Every paid tier includes monthly AI credits. When they run low, a $20 recharge adds 15,000 credits for frontier councils, web search, and Studio work — no personal API key, no separate provider account. 15,000 credits · $20 + tax
Free proves the method.
Paid sharpens the judgment.
Tries, plainly stated
The Free Council offers up to 20 asks per day. AI Model Switcher has a separate 10-message preview per visitor, including up to 3 messages each with Claude and Grok. Free previews depend on the shared daily allowance and provider availability. Paid plans include monthly AI credits and a separate daily model-call allowance; a council question can use several calls, so the number of complete sessions varies. Check your current allowance in the app. For paid plans, a $2.99 reset resets the daily model-call allowance, and a $20 recharge adds 15,000 AI credits for frontier councils and web search when the work needs live internet context.
Why it's built this way
A council is only as honest as its independence, and only as useful as your ability to actually convene it.
So every tier ships the complete ritual — nothing is held back but the caliber of the seats. The blind round is never skipped, the cross-examination is never softened, and the daily allowances are set where a serious person doing serious work will rarely feel them. When you do outgrow them, a recharge adds credits on the spot — no separate provider account, no key to manage.
The theory behind the ritual — why independent, diverse judgments outperform any single mind — is argued in full in The Council Thesis.