Pricing
A plan buys one thing here: how much you can poll in a day. It does not unlock a place, a feature, a bigger panel or a better model — the engine reads your plan in exactly one place, and that place is the daily allowance.
Today
Pro is not on sale yet.
The paid tier is built and the engine already knows how to honour it — what is missing is that this deployment does not yet put your plan on your sign-in, so a Pro subscription would bill you and change nothing about your day. We are not selling that. Everyone is on the free allowance below, us included. If you already hold a subscription — one started before we closed this, or set up for you by hand — sign in and the strip further down this page says what it is actually getting you, with a one-click way to stop paying for it.
Tell us you want it — support@populationtwins.comFree
Freeno card
Enough to answer a real question and check the answer's receipt, every day, without asking anyone for a budget.
Pro
$29.00per month
For the week where you are asking a question an hour rather than a question a day. Cancel from the billing portal; access runs to the end of the period you have paid for.
About 20 polls a day — on the deployment Helm runs.
We set the free allowance to 700,000 model tokens per account per UTC day, which is about 20 polls at roughly 35,000 tokens each, and about 200 on Pro. That is our configuration, not a property of the product: this page is served from the same code wherever the engine is run, and a deployment can set the allowance higher, lower, or not at all — unset is the default, and it turns the limit off entirely. Two things hold wherever you are reading this: Pro is 10× whatever the free allowance is, and the refusal you get on reaching one names it.
The allowance
It is metered in tokens.
The engine counts the model tokens a poll spends, not the polls themselves. A poll reads one archetype at a time, so a question that splits the population finely costs more than one that does not — which is why any figure stated in polls says “about”. A day of small questions goes further than a day of big ones.
Its size is set where the engine runs, not here.
How large the allowance is, and whether one is enforced at all, is a setting on the engine this site talks to. This page cannot read it, and it will not guess: the figure below is the one we run at, not a promise about wherever you are reading this. Signed in, you never have to take our word for it — the cost box above the ask prints your own allowance and what is left of it before you spend anything, and prints nothing where no allowance is enforced.
It resets at 00:00 UTC.
Not on a rolling window and not on your local midnight. Spend it all at 23:00 UTC and the next day starts an hour later.
Running out pauses what asks the model.
Polls, interviews, screener segments and reading an uploaded image answer “Daily poll allowance reached — resets at 00:00 UTC”, with the exact reset time attached. Building a population, forking a branch, browsing the atlas, and opening every result and export you already have are all outside the gate and keep working. A handful of other signed-in reads — your job handles, your saved artifacts — sit behind the same middleware and pause with them; none of them is a thing you would be in the middle of.
One allowance per account, not per browser.
It is counted against the account, not the session, so a second tab and a second machine draw on the same day rather than on a fresh one.
Building the population is free.
Compiling a place's synthetic population makes no model call and costs nothing on either plan — it is the asking that is metered, not the population. So is every split, dossier, export and re-read of a result you already have.
What Pro does not buy
The engine reads your plan in one function and uses it in one calculation: how many tokens you may spend before tomorrow. So there is no version of this product behind the paywall.
Tell us the place and the question, or book a call and we will run it with you.