Seat vs Usage Pricing Simulator

Every AI tooling negotiation eventually reaches the same question: flat price per seat, or pay for what you use? This simulator answers it with your numbers — team size, seat price, a token-price basis, your median developer's consumption — and one input most comparisons skip: the shape of the usage distribution. The verdict, crossover point and both bills recompute live, with a chart showing where the flat seat line crosses the rising usage curve.

The distribution is the whole game. With uniform usage, the comparison is trivial arithmetic: seat price versus median consumption times the blended token rate (we assume a 3:1 input-output split at rates verified 2026-06-11). But real teams are power-law: the simulator's realistic preset weights developer usage as 1/rank^1.2, which produces the familiar shape where the top developer burns 8× the median. Under that shape, the team's total bill is dominated by two or three people — and the "average usage" figure your vendor quotes describes nobody on your team.

That is why the p90 callout exists. Seat pricing socializes the outliers: the light users subsidize the agent-driven power users, which feels fine until the vendor reprices the whole contract around them. Usage pricing makes outliers visible and accountable, at the cost of forecastability — finance hates a bill that doubles because one team adopted a new workflow. The crossover number tells you which regime you are in today; the distribution tells you how fast that answer changes as agentic usage grows.

The unstated assumption in every seat-versus-usage debate is that you know your per-dev consumption, and almost nobody does — most teams know the invoice total and guess the rest. Measuring it changes the negotiation: with per-developer attribution you can price both models exactly, predict overage on blended plans, and cap the outliers instead of arguing about them. That measurement layer is what FORG provides, at a deliberately boring flat per-seat price — so the observability bill is the one number in this whole simulation that never surprises anyone.