A contract is a promise that you will spend a certain amount. A wallet is the amount you actually spent plus the amount you are ready to spend next. Voice AI has been sold on contracts and parts. The wallet is the argument that it should not be.
What the market actually charges
The per-minute sticker is almost never the price. The platform fee is the orchestration layer; LLM inference is a separate vendor, voice is a separate vendor, telephony is a separate vendor. Stack it and you are at two to four times the headline, with costs that all rise together when volume spikes. The full comparison against Retell and Vapi is here.
One balance, one blended rate
A V-Rep meters everything, voice in, voice out, model reasoning, telephony, platform, against a single wallet at one blended rate. No line items, no per-seat tax, no four-to-six invoices. You fund a balance and you spend it on exactly what happens on each interaction and nothing else. The real numbers and tiers are on the pricing page.
Why a wallet changes behavior, not just billing
A contract makes you defend a number you committed to before you knew the answer. A wallet makes the cost of the next call legible while you can still do something about it. Auto-recharge keeps the line running; a $10 minimum and no annual contract means you can start small and let usage, not procurement, decide the budget.
The inversion this unlocks
Once the unit cost is one honest number, something useful happens at scale: a V-Rep gets cheaper per interaction as volume rises, while a human line gets more expensive. That is not a billing detail. It is a strategy. It is what makes the demand loop work, and it is what Go iPower documented.