Quorum answers a question by putting it to several frontier models at once and having them deliberate before a synthesis pass writes the reply. A mode is your version of that argument: which engines take which seats, whether critique runs on every question or only the hard ones, how many rounds, and which model gets to judge.
If you know a subject well enough to know where a general-purpose AI answer goes wrong, you know something worth encoding. Building it takes a browser and an afternoon.
No waitlist, no application, no early-access list. A signed-in account is the whole requirement.
Every mode gets a page that says whether it works
Not a rating, not a review count. A blind, comparative test: your mode’s answers and a single leading model’s answers, anonymised and order-randomised, scored by a judge panel that does not know which is which. The result is public, on your mode’s own page, whether or not it flatters you.
Standard
Quorum’s default mode · efficacy report
Real measured result
Loading the live figure…
What you get for building one
Authorship
It is yours
A mode carries your name on the gallery card and on its own report page. Nobody rewrites its description into marketing copy — the ingredients shown are derived from the configuration, so they cannot overstate what it does.
Control
Real levers, not presets
Seats, engine pool, whether critique is always on or gated to hard questions, round depth, and which models judge and synthesise. If the panel argues badly about your subject, you are the one who can fix it.
Evidence
A number you did not write
Test it against real questions before publishing, then certify against a locked protocol — 45 deliberations spread evenly across light, medium and hard questions, each scored by three judges twice over. The config is fixed server-side, so nobody can tune their own score into existence. Certify again as you revise and the runs stack up, which is why a mode’s published sample size grows past any single run. The verdict is computed, not claimed. How the rig works.
How the money works, and where it stands
Published modes earn a share of real subscriber revenue, attributed by how much each mode is actually used and net of what it really cost to run. Here is the whole mechanism, and then the part most pages would leave out.
01
Usage is attributed
Each mode accrues weighted usage from real calls. Attribution is per mode, not per creator, so a mode that gets used is what earns.
02
45% of net margin, on traffic we can attribute
On API and MCP calls, which name your mode on the request, your share is 45% of net margin — gross attributed revenue less the real compute the mode consumed. A mode running three frontier seats costs more to run and earns on what is left. Subscriber traffic cannot be pinned to one mode the same way, so it settles from a shared pool split by measured use; that formula is not final and we have not published it, which you should weigh before building for subscriber reach rather than API use.
03
Stripe Connect, your account
Payouts go directly to a Stripe account you own. Funds do not sit in Quorum’s operating balance on their way to you.
04
60-day holdback
A period between a mode earning and the payout settling, which is what makes refunds and chargebacks resolvable without clawing money back from you.
The part worth knowing before you start
There are two published modes today and both are Quorum’s own. No independent maker has published one yet, and no Mode Maker payout has been made. The payment rails are real and tested; the demand for any particular mode is not proven, because there has not been one to prove it with.
Build one because you want the thing to exist and want your name on whether it works. If it earns, that is the mechanism above doing its job — not a forecast we are in a position to make.
What building one actually involves
Four steps in a browser, then a real test run before anything is public.
01
Brief and Panel
What the mode is for, and which engines sit in its seats.
02
Deliberation and Staff
Depth, critique policy, and who judges and synthesises.
03
Test
A real run against real questions. Priced before you commit; dry runs are free.
04
Publish
Enters review, gets a listing and a report page, and creates your payout account.