---
source_url: "https://www.fype.dev/?utm_source=openai"
title: "Fype | Decoupled Payments Orchestration Engine for India"
mirrored_at: 2026-08-08T01:00:58.538Z
host: www.fype.dev
cited_in_42a: true
mirror_canonical: "https://index.42a.ai/www.fype.dev/index__q__utm_source_openai"
---

> **Original source:** https://www.fype.dev/?utm_source=openai

Billing for AI companies

Meter usage. Cap spend. See revenue against cost, per account, live. On any payment processor.

Setup

Under 10 minutes

Processors

Bring your own

Problem

## Software used to be free to serve. AI isn't.

Every request costs you real money. On a flat price, your heaviest users become your most expensive ones. You find out from an invoice.

Blind spots

Your billing tools were built for a world where the next customer was free.

#### Margin per customer

Your billing knows what you charged. Not what it cost you.

#### When it goes negative

The cost lands weeks later. You already served them.

#### Caps that actually stop

A limit that warns after the fact is a receipt.

52%

Average AI gross margin

Classic software runs 70–90%.

ICONIQ, ~300 software execs

175×

What one power user cost

$35,000 of compute on a $200 plan.

Power-law usage analysis

$20

Lost per Copilot user, monthly

Priced at $10. Some users cost $80.

Wall Street Journal

Platform

## Billing that knows what your customers cost you.

Metering, credits, caps, entitlements and recovery in one layer. Margin on every screen. No processor lock-in.

Metering

### Meter anything.  
Price it on arrival.

Tokens, seconds, calls. Every event gets a price and your cost, on the same row.

usage.record()18.4k/min

gpt-5

18,412 tokens Northstar

$0.2120

claude-opus

7,088 tokens Vela AI

$0.1840

whisper

412 seconds Acme

$0.0247

embed-3

240,000 tokens Acme

$0.0192

a100-gpu

94 seconds Mira

$0.0862

Per account

Billed

$1,200

Cost

$475

Gross margin60.4%

Margin

### See who is profitable, right now.

Revenue and cost on the same row, per account.

Northstar Labs

+60%

Acme Cloud

+53%

Vela AI

−188%

Alerted today. Not four weeks from now.

Credits & caps

### A limit that refuses the request.

Caps enforced before the model call, not after.

Prepaid balance$12.40

spent $87.60cap $100.00

Blocked at the cap

402 credit\_limit\_reached

Entitlements

### One check, before every call.

await fype.can("agent\_runs")

true · 3ms · edge cached

Agent runs

412 / 2,000

Long context

unlimited

Fine tuning

upgrade

Repackage in the dashboard. Your code never changes.

Recovery

### Failed renewals, chased for you.

Smart retries that stop the moment the charge clears.

1instantsoft decline

2+2 daysretried, card updated

3+5 daysrecovered $480

Recovered this month

$18.2K

Portability

### The processor is a setting, not an architecture.

Your billing lives in Fype. Swap the rail underneath.

Setup

## One billing layer. Any processor.

Grab your keys and wire it up in minutes. Or hand the docs to your coding agent.

Developers

## Built by people who hit this wall themselves.

Meter an event in one line. Replay real traffic locally. The core is open source, so you can read what prices your revenue.

## Frequently asked questions

Wondering how this compares to your current billing stack? A founder answers.

[Contact Support](mailto:support@fype.dev)

It meters what your customers use, caps their spend before a request hits your model, and shows revenue against your cost for every account. You see margin today, not on next month's invoice.

Stripe knows what you charged. It has no idea what the customer cost you, because classic software had no cost per user. Fype takes in your cost of goods and reports margin per account.

They meter what you bill. None of them compare it against what you pay your model provider. That comparison is the product. We also handle GST invoicing and UPI mandates, which they don't.

No. Fype sits above it. Your catalog, subscriptions, credits and entitlements live with us. Stripe, Razorpay or Cashfree moves the money underneath, and swapping is a setting.

Single-digit milliseconds, edge cached. Your app asks one question before every request. Change your pricing in the dashboard and that question stays the same.

The request gets a 402 with a clear reason, before you spend money on it. Prefer a warning instead of a block? Switch it to soft overage.

Yes. We import your prices, subscriptions and usage, then run alongside your current setup while you check the numbers. Billing state lives with us, so moving in or out doesn't touch your app.