Balance and usage

The balance pays for AI features (the writing assistant, reply and compose drafts, Remie, contact research, AI variables, AI steps in campaigns and automations, the inbox agent and AI spam analysis), web searches those features make, and inbox placement tests past the monthly free tests. These features are paid-only: a free trial has no allowance, and the AI surfaces stay locked until the workspace is on a paid plan.

A dedicated reviewer workspace starts with a small test balance on its managed Test plan. It uses the normal balance and spend rules, and does not keep paid features unlocked after the reviewer grant expires. Joining an existing workspace as a tester does not add to its balance. See reviewer accounts.

The allowance resets to the plan amount at the start of each billing cycle, and the reset date is shown beside the balance. An annual plan still gets it every month: it resets on the same day of each month as the day the plan started (the last day of a shorter month when that day does not exist). Unused allowance is replaced by the fresh one and does not carry over, and once a month's allowance is used up, money you added pays for the rest.

Spending draws the allowance first and only touches money you added once the allowance is used up, so what you added is held back for heavy months.

A balance held in credits before SenditCold switched to dollars was converted to dollars, and purchased credits carry over as money you added, so they still never expire.

Every charge is the actual cost of the work:

• AI calls are priced per token on the model that ran them, at the provider's price per million input and output tokens.
• Other units have a price each: a web search an AI step makes, a paid placement test, or lead finder work (a website read, a person's email found on request, a search page through a source the instance pays for).
• The markup is added on top of the provider price. It is 25% by default, so each $1.00 of provider cost is charged as $1.25. The operator of the instance sets the price list and the markup.

At the default prices, a standard model costs $3.75 per million input tokens and $18.75 per million output tokens once the markup is added, and a lighter model costs less. The default OpenAI model, gpt-6-luna, costs $0.125 per million input tokens and $0.625 per million output tokens with the markup, so on it each action below costs a small fraction of what the table shows. What typical actions cost at the default prices on a standard model:

Longer threads, bigger prompts and longer outputs cost more, and shorter ones cost less. The history shows exactly what each call used.

Rules for every charge:

• Before work starts, the balance has to cover a typical call of that kind. A short balance is refused before any work starts, so you are never half-charged.
• After the call, the actual cost is charged. A result you already received never fails for costing more than expected; at worst the balance goes to zero.
• A provider error charges nothing, so you never pay for output you did not get.
• A contact research run that finds nothing is still charged for the searching it did.
• A free or local model (on a self-hosted instance) is never charged.

The balance also pays for inbox placement tests once a paid workspace has used its free tests for the month. A test is a flat price ($0.50 at the default prices, set by the operator), with no AI involved. Nothing is charged without you agreeing to it: the new-test dialog shows the price and your balance, and you tick to pay before the test can start. A test that delivers no copy is refunded to the part of the balance it came from. Placement charges count toward your spend limits (a refunded one no longer does) and appear in the usage breakdown as Placement tests; they are not held to the AI generation caps. Tests on your own seed inboxes are always free.

Every charge, addition and reset, newest first. A charge shows the activity, the model, the input and output tokens or the units used, the amount in dollars, the balance after it, and the time. Each one records exactly what ran it: campaign, step and contact for a switch step; automation, node and run for an automation step (including what an Ask AI branch asked); the thread for inbox drafts; and the teammate behind manual features.

The usage card above it shows spend today, this week and this month, a graph over the last 7, 30 or 90 days, and breakdowns by feature and by model (dollars, tokens and call counts).

You can add any amount from $5 to $5,000, in whole cents. Adding money needs an active paid plan.

• One click on the saved card. The card is charged at once and the money is added as soon as the payment goes through. If the payment is still going through when the page answers, the Balance tab says so and updates when it lands; SenditCold finishes it in the background, and another payment cannot start until it does. Each charge is billed as a Stripe invoice, so it has its own invoice and receipt.
• Pay another way. Opens Stripe Checkout for the amount. Use it when there is no saved card, when you want another card, or when the saved card is declined or your bank asks to confirm the payment. Checkout saves the card for next time. Money arrives only after Stripe confirms the payment, so closing the checkout tab charges nothing, and a completed payment is added exactly once even if Stripe retries.
• Change the card. Opens Stripe Checkout to save a new card, without charging anything.

The card shown is the one on the workspace's Stripe billing account (brand and last four digits). Card details are held by Stripe and never touch SenditCold.

Automatic reload adds money on the saved card when the balance falls below a threshold, so AI never pauses for an empty balance. It needs a saved card and an active paid plan.

The monthly limit counts every automatic reload in the calendar month. When less than a full reload is left under it, SenditCold adds what is left if that is at least $5, and otherwise waits for next month.

When a reload is declined:

• A temporary decline (for example insufficient funds) is retried 1 hour, 6 hours and 18 hours later. If the last retry also fails, automatic reload turns off.
• A hard decline, or your bank asking to confirm the payment, turns automatic reload off at once, so nothing else is charged. Everyone who manages billing is notified with the reason. Add money once (through Checkout if the bank needs you to confirm), update the card if it was declined, then turn automatic reload back on.
• Stripe keeps failing to answer: the reload is given up after 8 tries and automatic reload turns off the same way.

A reload waiting for a retry is canceled, never charged, when you turn automatic reload off or the balance is back above the threshold by the time it is due.

The Balance tab shows the last reload, how much has been reloaded this month, and a pending retry with its time.

Everyone with Manage billing gets an in-app notice when the balance falls below your low balance threshold ($0.50 by default, set on the Balance tab), and another when it runs out. Each one is sent at most once a month, and again after money is added. Neither is sent when automatic reload will top the balance up.

Each limit shows current spend against it, and changes collect into a save bar. Who can use AI at all is separate: the Use AI role permission (see Team and roles).

Every payment raises a Stripe invoice: each one-click addition, each automatic reload, each Checkout payment, and the subscription itself. The Balance tab lists the latest ones with a link to each invoice and its PDF. Invoices and receipts for the plan itself are covered in Plans and billing.

Does unused allowance roll over? No. Money you add carries over.

Does money I add expire? No, through any number of resets.

What happens when I change plans? Your allowance follows the new plan from the next cycle. Money you added is unaffected.

What happens if I cancel? The monthly allowance stops; money you added stays until spent, and is usable again once the workspace is on a paid plan.

Does the balance apply when self-hosting? Not when billing is off, which is the self-host default (BILLING_PROVIDER=none). You pay your AI provider directly, so the balance is bypassed: nothing is charged, there is no plan allowance, the header shows no balance, and per-action cost hints are hidden. A self-hosted instance that runs Stripe (BILLING_PROVIDER=stripe) is metered exactly like the hosted product, at the price list and markup its operator sets (see instance settings). Separately, a free or local model (AI_PROVIDER=ollama, or any provider with AI_FREE=true) is never charged and shows a "free model" notice. See the self-hosting guide.