Official Velobase Billing SDK for JavaScript and TypeScript.
Velobase is a customer-aware AI gateway: you route your app's LLM traffic through it, and every call is metered and billed against the end customer's credits wallet automatically. No token counting, no cost math, no settlement code on your side.
This SDK covers everything around that gateway:
- Fund wallets —
customers.deposit()when your customer pays you - Check balances —
customers.get()for display and paywalls - Bill non-LLM work —
billing.freeze/consume/unfreeze/deductfor media jobs, per-task charges, manual adjustments - Reconcile —
usage.list()andreceipts.get()over gateway-billed LLM traffic - Handle paywalls —
parseGatewayInsufficientBalance()for typed 402 handling
The SDK does not wrap LLM calls. Your existing LLM SDK talks to the gateway directly; see Point LLM traffic at the Velobase gateway.
- Zero runtime dependencies — uses native
fetch - Works with Node.js 18+, Deno, Bun, and Cloudflare Workers
- ESM and CommonJS dual build with full TypeScript declarations
- Automatic retries with exponential backoff
npm install @velobaseai/billing
# or
pnpm add @velobaseai/billing
# or
yarn add @velobaseai/billingimport Velobase, { GATEWAY_HEADERS } from '@velobaseai/billing';
import { createOpenAI } from '@ai-sdk/openai';
const vb = new Velobase({ apiKey: 'vb_live_xxx' });
// 1. Your customer pays you — fund their wallet (creates the customer if new)
await vb.customers.deposit({
customerId: 'user_123',
amount: 5_000_000, // $5.00; 1 USD = 1,000,000 credits
});
// 2. Route LLM traffic through the gateway. Each call is metered and
// billed against user_123's wallet automatically.
const openai = createOpenAI({
baseURL: 'https://api.velobase.io/v1',
apiKey: 'vb_live_xxx',
headers: { [GATEWAY_HEADERS.customer]: 'user_123' },
});
// 3. Show the balance
const customer = await vb.customers.get('user_123');
console.log(customer.wallets.default.available); // credits remainingThere are two billing paths, and the split is by workload:
LLM calls: implicit, at the gateway. Point your LLM SDK at the Velobase
gateway with an X-Velobase-Customer header. The gateway freezes an estimate
before forwarding upstream, settles the real token cost when the response
finishes, and refunds on failure. You never call freeze or consume for LLM
traffic; doing so would double-charge the customer.
Everything else: explicit, through this SDK. Media generation, per-task charges, manual adjustments — anything that is not an LLM call through the gateway — uses the ledger primitives directly:
deposit → freeze → consume (normal flow)
→ unfreeze (failure/cancellation)
deposit → deduct (immediate deduction)
Both paths drive the same per-customer ledger. Implicit gateway billing is the same freeze → consume/unfreeze cycle, executed by the gateway on your behalf; gateway-billed LLM calls and your explicit charges land in the same wallets and show up in the same ledger, usage, and receipt queries. The primitives are the foundation, not a legacy path.
- Deposit — Add raw credits to a customer's wallet. Creates the customer automatically on first deposit. 1 USD = 1,000,000 credits. Supports
wallet/sourcefor balance categories, andstartsAt/expiresAtfor time-limited credits. - Freeze — Pre-authorize a credits amount before performing work. The frozen credits are deducted from
availablebut not yetused. Each freeze is identified by a uniquetransactionIdyou provide. - Consume — After the work is done, settle the frozen amount. You can pass
actualAmountto charge less than what was frozen; the difference is automatically returned. - Unfreeze — If the work fails or is cancelled, release the full frozen amount back to the customer.
- Deduct — Directly deduct credits from a customer without freezing first. Useful for immediate charges.
- Ledger — Query a customer's transaction history with filtering and cursor-based pagination.
- Gateway usage / receipts — Query LLM gateway usage rows and per-transaction receipts for reconciliation.
All write operations are idempotent — repeating the same transactionId (freeze/consume/unfreeze/deduct) or idempotencyKey (deposit) returns the original result without double-charging. Gateway LLM requests accept an Idempotency-Key header with the same guarantee.
const vb = new Velobase({
apiKey: 'vb_live_xxx', // Required. Your Velobase API key.
baseUrl: 'https://api.velobase.io', // Optional. Override the API endpoint.
timeout: 30000, // Optional. Request timeout in ms (default: 30s).
maxRetries: 2, // Optional. Retry count on 5xx/network errors (default: 2).
});When baseUrl is omitted, the SDK resolves the endpoint as: explicit
baseUrl option > VELOBASE_BASE_URL environment variable >
https://api.velobase.io. SDK defaults are always production-safe; published
releases never bake in a non-production URL.
- Production — omit
baseUrl(or pinhttps://api.velobase.io). Use the stable npm release. - Development / staging — set
VELOBASE_BASE_URLin the deployment's environment (or passbaseUrlexplicitly), and keep its API key separate from production. A staging deployment must never point at the production ledger: it would bill real customer wallets. Guard for this at startup rather than relying on configuration discipline. - The gateway (LLM traffic) and the control plane (this SDK) accept the same project API key and, in standard deployments, the same host. One key, one host, both planes.
The billing SDK does not wrap LLM calls. Use your LLM SDK with the Velobase gateway base URL, pass your project API key as the upstream API key, and attach the customer/trace headers for implicit per-call billing.
import { createAnthropic } from '@ai-sdk/anthropic';
import { createOpenAI } from '@ai-sdk/openai';
import Velobase, { GATEWAY_HEADERS } from '@velobaseai/billing';
const gatewayUrl = 'https://api.velobase.io';
const gatewayApiKey = process.env.VELOBASE_GATEWAY_API_KEY!;
const customerId = 'user_123';
const traceId = 'turn_abc';
const anthropic = createAnthropic({
baseURL: `${gatewayUrl}/v1`,
apiKey: gatewayApiKey,
headers: {
[GATEWAY_HEADERS.customer]: customerId,
[GATEWAY_HEADERS.traceId]: traceId,
},
});
const openai = createOpenAI({
baseURL: `${gatewayUrl}/v1`,
apiKey: gatewayApiKey,
headers: {
[GATEWAY_HEADERS.customer]: customerId,
[GATEWAY_HEADERS.traceId]: traceId,
},
});
const vb = new Velobase({ apiKey: gatewayApiKey, baseUrl: gatewayUrl });
const usage = await vb.usage.list({ traceId });Billing at the gateway is implicit: an estimate is frozen before the request goes upstream, the real token cost settles when the response completes, and failed requests are refunded. Do not freeze or consume for these calls.
For retried LLM requests, send an Idempotency-Key header per logical
request: a replayed request returns a 409 instead of charging twice.
When a customer's wallet cannot cover the estimated cost, the gateway rejects
the call with HTTP 402 before any tokens are spent. parseGatewayInsufficientBalance()
extracts the typed paywall payload from whatever error shape your LLM SDK
throws:
import { parseGatewayInsufficientBalance } from '@velobaseai/billing';
try {
await streamText({ model, messages });
} catch (err) {
const paywall = parseGatewayInsufficientBalance(err);
if (paywall) {
// Show your top-up UI. All amounts are credits (1 USD = 1,000,000 credits).
console.log(paywall.requiredCredits); // estimated cost of the rejected call
console.log(paywall.availableCredits); // what the customer has left
console.log(paywall.requiredUsd); // same, pre-formatted in USD
} else {
throw err;
}
}Tip:
requiredCreditsreflects the gateway's pre-call hold. Its output part is sized from the request'smax_tokens; a request that omitsmax_tokensreserves for a large 8,000-token output and gets a much higher paywall threshold than the call will actually cost. Always setmax_tokenson gateway calls. Image and PDF inputs are supported on gateway chat calls, and their hold is estimated from image dimensions and PDF page count, so media requests keep sane thresholds too.
// Safe to retry — the second call returns the same result without double-charging
const result = await vb.customers.deposit({
customerId: 'user_123',
amount: 5_000_000, // $5.00; 1 USD = 1,000,000 credits
idempotencyKey: 'order_abc_payment',
description: 'Purchase, $5.00',
});
console.log(result.addedAmount); // 5000000
console.log(result.isIdempotentReplay); // false on first call, true on retriesconst result = await vb.customers.deposit({
customerId: 'user_123',
amount: 10_000_000, // $10.00; 1 USD = 1,000,000 credits
wallet: 'bonus',
source: 'annual_campaign',
startsAt: '2025-01-01T00:00:00Z',
expiresAt: '2025-12-31T23:59:59Z',
description: 'Annual bonus, $10.00',
});
console.log(result.wallet); // 'bonus'
console.log(result.source); // 'annual_campaign'
console.log(result.startsAt); // '2025-01-01T00:00:00.000Z'
console.log(result.expiresAt); // '2025-12-31T23:59:59.000Z'const result = await vb.customers.deposit({
customerId: 'user_123',
amount: 10_000_000,
name: 'Alice',
email: 'alice@example.com',
metadata: { plan: 'pro', source: 'stripe' },
});Use the ledger primitives for work that does not go through the gateway: media generation, long-running jobs, one-off charges.
const CUSTOMER = 'user_123';
// Generate transactionId once and persist before freezing
const transactionId = `${CUSTOMER}_${crypto.randomUUID().replace(/-/g, '')}`;
// Check balance before starting
const before = await vb.customers.get(CUSTOMER);
console.log('Available:', before.wallets.default.available);
// Freeze the estimated cost
await vb.billing.freeze({
customerId: CUSTOMER,
amount: 100,
transactionId,
businessType: 'TASK',
description: '1080p video, ~60s',
});
// ... do the work ...
// Settle with the actual cost (partial consumption)
const result = await vb.billing.consume({ transactionId, actualAmount: 73 });
console.log('Charged:', result.consumedAmount); // 73
console.log('Returned:', result.returnedAmount); // 27
// Verify final balance
const after = await vb.customers.get(CUSTOMER);
console.log('Available:', after.wallets.default.available);const CUSTOMER = 'user_123';
const transactionId = 'api_call_001';
const result = await vb.billing.deduct({
customerId: CUSTOMER,
amount: 5,
transactionId,
businessType: 'TASK',
description: 'API call charge',
});
console.log('Deducted:', result.deductedAmount); // 5
console.log('At:', result.deductedAt);const usage = await vb.usage.list({
customerId: 'user_123',
traceId: 'turn_abc',
});
for (const row of usage.data) {
console.log(row.transactionId, row.customerCostCredits, row.status);
}
const receipt = await vb.receipts.get(usage.data[0].transactionId);
console.log(receipt.cost.credits, receipt.balance.credits);// List all ledger entries (default limit=20)
const ledger = await vb.customers.ledger('user_123');
for (const entry of ledger.items) {
console.log(entry.operationType, entry.amount, entry.wallet, entry.source, entry.createdAt);
}
console.log('Total:', ledger.totalCount);
// Filter by operation type
const grants = await vb.customers.ledger('user_123', { operationType: 'GRANT' });
// Filter by transactionId
const txnEntries = await vb.customers.ledger('user_123', { transactionId: 'job_abc' });
// Paginate with cursor
const page1 = await vb.customers.ledger('user_123', { limit: 10 });
if (page1.hasMore) {
const page2 = await vb.customers.ledger('user_123', { limit: 10, cursor: page1.nextCursor });
}const customer = await vb.customers.get('user_123');
// Wallets are keyed by wallet/category name.
const defaultWallet = customer.wallets.default;
defaultWallet.total; // total deposited
defaultWallet.used; // total consumed
defaultWallet.frozen; // currently frozen (pending)
defaultWallet.available; // total - used - frozen
// Sources inside each wallet preserve validity windows.
for (const source of defaultWallet.sources) {
console.log(source.source); // 'default', 'stripe_checkout', etc.
console.log(source.available);
console.log(source.startsAt); // null or ISO date string
console.log(source.expiresAt); // null or ISO date string
}Deposit funds. Creates the customer if they don't exist.
| Parameter | Type | Required | Description |
|---|---|---|---|
customerId |
string |
Yes | Your unique customer identifier |
amount |
number |
Yes | Raw credits to deposit. Must be a positive integer. 1 USD = 1,000,000 credits. |
wallet |
string |
No | Wallet/category to credit. Defaults to "default" on server. |
source |
string |
No | Optional source label for this grant. Defaults to "default" on server. |
startsAt |
string |
No | ISO datetime string. When the credits become active. |
expiresAt |
string |
No | ISO datetime string. When the credits expire. Must be after startsAt. |
idempotencyKey |
string |
No | Prevents duplicate deposits on retry |
name |
string | null |
No | Customer display name |
email |
string | null |
No | Customer email |
metadata |
object |
No | Arbitrary key-value metadata |
description |
string |
No | Description for the deposit |
Returns: { customerId, accountId, wallet, source, totalAmount, addedAmount, startsAt, expiresAt, recordId, isIdempotentReplay }
Retrieve a customer's balance and account details.
Returns: { id, name, email, metadata, balance, accounts, createdAt }
Query a customer's transaction history with filtering and cursor-based pagination.
| Parameter | Type | Required | Description |
|---|---|---|---|
customerId |
string |
Yes | Customer identifier (positional) |
params.limit |
number |
No | Page size (1–100, default 20) |
params.cursor |
string |
No | Cursor from a previous nextCursor for pagination |
params.operationType |
string |
No | Filter by operation type: FREEZE, CONSUME, UNFREEZE, GRANT, EXPIRE |
params.transactionId |
string |
No | Filter by transaction ID |
Returns: { items: LedgerEntry[], totalCount, hasMore, nextCursor }
Each LedgerEntry has: id, operationType, amount, wallet, source, transactionId, businessType, description, accountId, status, createdAt
Freeze credits before performing work.
| Parameter | Type | Required | Description |
|---|---|---|---|
customerId |
string |
Yes | Customer identifier |
amount |
number |
Yes | Credits to freeze (must be > 0). 1 credit = 1 micro-USD. |
transactionId |
string |
Yes | Your unique ID for this operation (idempotency key) |
businessType |
BusinessType |
No | Business category. See businessType for accepted values. |
description |
string |
No | Human-readable description |
Returns: { transactionId, frozenAmount, freezeDetails, isIdempotentReplay }
Settle a frozen amount. Supports partial consumption.
| Parameter | Type | Required | Description |
|---|---|---|---|
transactionId |
string |
Yes | The transactionId from the freeze |
actualAmount |
number |
No | Actual amount to charge. Defaults to full frozen amount. |
Returns: { transactionId, consumedAmount, returnedAmount, consumeDetails, consumedAt, isIdempotentReplay }
Release a frozen amount back to the customer.
| Parameter | Type | Required | Description |
|---|---|---|---|
transactionId |
string |
Yes | The transactionId from the freeze |
Returns: { transactionId, unfrozenAmount, unfreezeDetails, unfrozenAt, isIdempotentReplay }
Directly deduct credits without freezing first. Useful for immediate charges.
| Parameter | Type | Required | Description |
|---|---|---|---|
customerId |
string |
Yes | Customer identifier |
amount |
number |
Yes | Credits to deduct (must be > 0). 1 credit = 1 micro-USD. |
transactionId |
string |
Yes | Your unique ID for this operation (idempotency key) |
businessType |
BusinessType |
No | Business category. See businessType for accepted values. |
description |
string |
No | Human-readable description |
Returns: { transactionId, deductedAmount, deductDetails, deductedAt, isIdempotentReplay }
List LLM gateway usage rows for the current project. This endpoint is intended for reconciliation and trace-level reporting after LLM traffic has gone through the Velobase gateway.
| Parameter | Type | Required | Description |
|---|---|---|---|
customerId |
string |
No | Filter by X-Velobase-Customer |
model |
string |
No | Filter by public gateway model id |
transactionId |
string |
No | Filter by gateway transaction id |
traceId |
string |
No | Filter by X-Velobase-Trace-Id |
cursor |
string |
No | Cursor from a previous nextCursor |
limit |
number |
No | Page size |
Returns: { object: "list", data: UsageListItem[], hasMore, nextCursor }
Retrieve one LLM gateway billing receipt by transaction id.
Returns: { object: "billing.receipt", transactionId, status, customerId, traceId, model, usage, cost, frozen, balance, upstream, created }
Header-name constants for pointing LLM traffic at the gateway:
| Constant | Header | Purpose |
|---|---|---|
GATEWAY_HEADERS.customer |
X-Velobase-Customer |
Which end customer to bill. Required on billable gateway calls made with a project key. |
GATEWAY_HEADERS.traceId |
X-Velobase-Trace-Id |
Optional correlation id. Query it back with usage.list({ traceId }). |
Extract the typed 402 paywall payload from any error thrown by an LLM SDK
whose request was rejected by the gateway for insufficient balance. Returns
null when the error is anything else.
Returns: { requiredCredits, availableCredits, estimatedCredits, requiredUsd, availableUsd }
transactionId uniquely identifies one billing operation (freeze → consume/unfreeze cycle, or a single deduct) and acts as its idempotency key. The server uses it to prevent double-charging on retries.
Recommended format: {customerId}_{uuid}
// Generate once per billing operation, then persist it
const transactionId = `${customerId}_${crypto.randomUUID().replace(/-/g, '')}`;
// e.g. "user_123_a3f8c21d4e0b4a9f8c1d2e3f4a5b6c7d"Rules:
- Generate once and store — create the ID before calling
freeze()ordeduct(), save it to your database, and reuse the same value on retries - Never regenerate at the call site — calling
crypto.randomUUID()insidefreeze()produces a different ID on every attempt, breaking idempotency - Unique within your project — two different billing operations must not share the same
transactionId
// Wrong — new UUID on every call, idempotency broken on retry
await vb.billing.freeze({
customerId,
amount: 50,
transactionId: `${customerId}_${crypto.randomUUID()}`, // ❌ regenerated each time
});
// Correct — UUID generated once and persisted before calling freeze
const transactionId = await db.getOrCreateTransactionId(operationId, customerId);
// e.g. returns existing ID or stores `${customerId}_${crypto.randomUUID()}` on first call
await vb.billing.freeze({ customerId, amount: 50, transactionId });
// Safe to retry — same transactionId returns the original result
await vb.billing.freeze({ customerId, amount: 50, transactionId });businessType is an optional field on freeze() and deduct() that categorises the billing operation for analytics and reconciliation. The SDK validates the value client-side before sending the request.
Accepted values:
| Value | Description |
|---|---|
UNDEFINED |
Default / unclassified (server default when omitted) |
TASK |
Async task execution (e.g. video generation, image processing) |
ORDER |
One-time purchase or order fulfilment |
MEMBERSHIP |
Membership plan credit grant |
SUBSCRIPTION |
Subscription renewal credit grant |
FREE_TRIAL |
Free-trial credit grant |
ADMIN_GRANT |
Manually granted credits by an admin |
Passing an unrecognised value throws an Error immediately — before any network request is made.
import Velobase, { type BusinessType } from '@velobaseai/billing';
const vb = new Velobase({ apiKey: 'vb_live_xxx' });
await vb.billing.freeze({
customerId: 'user_123',
amount: 50,
transactionId: 'job_abc',
businessType: 'TASK', // ✅ IDE autocomplete + client-side validation
});
await vb.billing.freeze({
customerId: 'user_123',
amount: 50,
transactionId: 'job_abc',
businessType: 'INVALID_VAL', // ❌ throws Error before making a network call
});All API errors throw typed exceptions that extend VelobaseError:
import {
VelobaseError,
VelobaseAuthenticationError,
VelobaseValidationError,
VelobaseNotFoundError,
} from '@velobaseai/billing';
try {
await vb.billing.freeze({
customerId: 'user_123',
amount: 999999,
transactionId: 'job_xyz',
});
} catch (err) {
if (err instanceof VelobaseValidationError) {
// 400 — bad request or insufficient balance
console.error(err.message); // "insufficient balance"
} else if (err instanceof VelobaseAuthenticationError) {
// 401 — invalid or missing API key
} else if (err instanceof VelobaseNotFoundError) {
// 404 — customer not found
} else if (err instanceof VelobaseError) {
// catch-all for other API errors
console.error(err.status, err.type, err.message);
}
}| Error Class | HTTP Status | When |
|---|---|---|
VelobaseAuthenticationError |
401 | Invalid or missing API key |
VelobaseValidationError |
400 | Bad params, insufficient balance |
VelobaseNotFoundError |
404 | Customer or resource not found |
VelobaseConflictError |
409 | Conflicting operation |
VelobaseInternalError |
500 | Server-side error (auto-retried) |
The SDK automatically retries on 5xx errors and network failures with exponential backoff (500ms, 1s, 2s..., capped at 5s). Retries are safe because all Velobase write operations are idempotent.
4xx errors (validation, auth, not found) are never retried.
const { Velobase } = require('@velobaseai/billing');
const vb = new Velobase({ apiKey: 'vb_live_xxx' });latest— stable releases. Pin these in production (npm install @velobaseai/billing).dev— prerelease builds (x.y.z-dev.N) for development branches (npm install @velobaseai/billing@dev). Never ship a-devbuild to production.
SDK defaults are always production-safe: the default baseUrl points at the
production API in every release, on both channels.
MIT