# Budgets and receipts

Understand holds, micro-units, readiness-gated charges, and your itemized bill.

A rental budget is a spending cap, not a promise that the entire amount will be charged. The service holds funded budget, meters accepted runtime after verified readiness, and reports the resulting charge through billing and a receipt.

## Use integer money values

USDC and USDT amounts use decimal **strings** of micro-units in the API:

```text
"1000000" = 1 unit of the selected stablecoin
"2500000" = 2.5 units
```

Use `BigInt` or a decimal library for calculations. Do not convert monetary strings to binary floating-point numbers and send rounded results back to the API. Keep the chosen asset with every amount.

## Read balances by purpose

| Balance | Meaning for your workflow |
| --- | --- |
| `spendable_micro` | Credit available to fund new work |
| `held_micro` / `active_hold_micro` | Credit held by live rentals; these describe the same live-job hold |
| `pending_micro` | Pending transfer or withdrawal amounts, not new spending capacity |
| `confirming_micro` | Credit still subject to finality restrictions |
| `disputed_micro` | Amount under an open dispute hold |
| `withdrawable_micro` | Credit currently eligible for withdrawal checks |

Read the service's current balances rather than maintaining a competing client-side ledger.

## Review rate, cap, and fee

`machine_rate_micro_per_hour` is the rate for the whole machine. A quote reports `estimated_total_micro`, `budget_micro`, and `platform_fee_bps`. The current published platform fee is 10%, included in the listed charge rather than added on top; the quote freezes the actual applicable basis points.

The funded budget caps the rental's charge. An explicit budget extension is a new authorization that adds funds; a client must not silently extend because a task ran longer than expected. Metering begins at verified readiness. Use the receipt for the actual accepted runtime and fee split.

## Read the receipt

After stopping, request `GET /api/v1/billing/receipts/{job_id}`. Important fields include `accepted_runtime_ms`, `charge_micro`, `fee_micro`, `host_micro`, `funded_budget_micro`, `platform_fee_bps`, and journal `entries`.

Compare the receipt to the rental ID and chosen asset. Keep a redacted record for your task accounting. If the result is unexpected, use [support](https://docs.amplerun.com/support.md) with the rental and request IDs; do not publish account credentials or a full unredacted access response.

---
Last reviewed: 2026-10-19.
Canonical: https://docs.amplerun.com/billing/budgets-and-receipts/