A DimeBox wallet needs funding before it can authenticate any paid call, and there are two ways to put money into one: a one-time top-up through Stripe Checkout whenever you need it, or a monthly subscription that recredits the wallet automatically every period. Both land on the same balance, spent down at the same prices through the same Authorization: Bearer <wallet_secret> header. The difference is entirely in how the balance gets there, and what that costs over time.
A one-time top-up is exactly what it sounds like: a card charge through Stripe Checkout that credits the wallet once. Do it again whenever the balance runs low, or pair it with auto-reload so it happens automatically at a threshold you set - see Auto-Reload for Agent Wallets. There's no ongoing commitment; stop topping up and the wallet just stops getting funded, with whatever balance it has left.
A monthly subscription recredits the wallet automatically every billing period, and - notably - skips the one-time setup fee that a normally created wallet otherwise carries. Subscriptions for Agent Wallets goes into why that trade exists. The cost is a standing recurring charge that keeps running until you cancel it, whether or not that period's credit actually gets spent.
It comes down to how predictable your call volume is. A wallet that gets topped up sporadically - a side project, an agent that runs occasionally, usage that varies a lot month to month - pays the setup fee once and otherwise only pays for what it actually tops up, with no risk of paying for a subscription period it doesn't use. A wallet behind steady, predictable daily volume amortizes the subscription's avoided setup fee over enough calls that the recurring commitment stops mattering, and the "it just refills itself" property removes one more thing to remember.
It's worth being precise about what the subscription actually subscribes you to: a recurring wallet top-up, not a recurring plan for calling DimeBox. Every call is still metered and priced per call - $0.02 to draft, $0.01 to send, $0.03 for priority receive - whether the balance behind it came from a one-time charge or a subscription credit. Pay-Per-Call vs Subscriptions and No Subscriptions, No Seats cover why the core API stays metered regardless of how you choose to fund the wallet that pays for it.
Nothing locks you into one path permanently. A wallet started with one-time top-ups can move to a subscription later if volume grows into predictable; a subscribed wallet can cancel and fall back to manual or auto-reloaded top-ups if volume drops. The ledger underneath doesn't care which funding method produced a given credit - see The Full Lifecycle of a DimeBox Wallet for how both funding paths land in the same place.
Most email and API tooling bills you a flat monthly fee regardless of usage, with a seat count or a call-volume tier attached. DimeBox's subscription option isn't that - it funds a balance that still gets metered per call, not a plan that unlocks a volume tier. Subscription Email Tools vs Pay-Per-Call covers that broader contrast if you're coming from a background of flat-fee SaaS pricing and trying to map this onto it.
If you're not sure which fits, start with one-time top-ups and auto-reload - it's the lower-commitment option, and it's straightforward to see from your actual DimeBox call volume over a few weeks whether a subscription would have been cheaper. Configure either one from /wallet.
The math is simple once you know your own call volume. If the one-time setup fee is small relative to a month of calls, skipping it via a subscription barely moves the needle - you're trading a single upfront cost for a recurring one of roughly the same size, spread out instead of paid once. The subscription starts winning once you're confident you'll keep using the wallet past the first billing period or two, because at that point the setup fee you'd have paid once under the one-time plan is already smaller than the recurring charges you'd have accumulated anyway. It starts losing the moment usage drops to occasional or stops altogether, because a subscription keeps charging regardless.
There's no penalty for picking the plan that turns out not to fit beyond paying slightly more than necessary in the meantime. A one-time wallet that ends up needing monthly refills can switch to a subscription; a subscribed wallet that ends up used rarely can cancel. The cost of guessing wrong is a few dollars of friction, not a locked-in decision, which is why it's reasonable to start with whichever plan matches your best guess at usage and adjust once you have real data.
Default to one-time top-ups when you're not sure, since switching from one-time to a subscription later is simpler than untangling a recurring charge you signed up for and forgot about. Auto-reload closes most of the convenience gap between the two options anyway, which is why the safer default for a sporadically used wallet is one-time funding with a conservative auto-reload threshold, not a subscription signed up for on day one.