# Billing

The [webforai platform](https://webforai.dev/platform) is metered in **credits**. Every operation reports a credit amount; nothing else is counted — no per-seat pricing, no request tiers, no minimum commitment.

## Credit schedule

| operation | credits |
| - | - |
| scrape via `fetch` | 1 |
| scrape via `browser` | 2 |
| scrape via `proxy-fetch` (subscription only) | 2 |
| scrape via `proxy-browser` (subscription only) | 3 |
| `screenshot` option | +1 |
| `rehostImages` option | +1 per started 5 images |

The default [`auto` engine](https://webforai.dev/platform/api-reference#the-auto-engine) has no price of its own: the operation bills the engine that actually produced the result (so 1 credit when a plain fetch sufficed, 2 when the page needed browser rendering), and only that engine.

Batch and crawl jobs bill **per successfully converted page** using the same schedule. A page that fails is not billed, and neither is a request that errors before the conversion runs.

## Price

- **1,000 credits per month free** — no subscription, no card.
- Beyond that, a metered monthly subscription with graduated tiers per calendar month:

| credits in the month | price per credit | per 1,000 `fetch` pages | per 1,000 `browser` pages |
| - | - | - | - |
| first 1,000 | free | free | free |
| 1,001 – 100,000 | **\$0.001** | \$1.00 | \$2.00 |
| 100,001 – 1,000,000 | \$0.0007 | \$0.70 | \$1.40 |
| above 1,000,000 | \$0.0005 | \$0.50 | \$1.00 |

Months are calendar months in UTC: the free allowance, the tiers and the spend cap all reset on the 1st at 00:00 UTC. Each tier applies only to the credits inside it. 1,000 free credits is 1,000 `fetch` scrapes, or 500 browser-rendered or `proxy-fetch` pages, every month. For example, 50,000 `auto` pages a month where 20% need a browser is 60,000 credits: 1,000 free + 59,000 × \$0.001 = **\$59**.

**Price change (2026-09-24).** Prices went down: `browser` from 5 to 2 credits, `proxy-browser` from 5 to 3, the per-credit price from \$0.002 to graduated tiers starting at \$0.001, and the free allowance from 500 to 1,000 credits a month. Routing through the proxy adds one credit to the engine (`fetch` 1 → `proxy-fetch` 2, `browser` 2 → `proxy-browser` 3).

## Proxy engines need a subscription

`proxy-fetch` and `proxy-browser` — and `auto` with a non-`auto` `region` such as `"jp"`, which runs on them — are available with an active subscription only. On the free allowance they are refused with `402 payment_required` before anything runs, and nothing is billed. The free allowance covers `fetch`, `browser` and `auto`. The public demo never uses the proxy.

## Spend cap

Every subscription has a monthly **spend cap**: **\$50 by default**, adjustable on the [dashboard](https://platform.webforai.dev/dashboard) from \$1 to \$5,000 (there is no "unlimited"). When this calendar month's estimated bill reaches the cap, further requests return `402 spend_cap_reached` until you raise the cap or the month ends, and a running batch or crawl job stops instead of converting further pages. The cap is checked before each operation against usage already recorded, so requests that are in flight at the same moment can each finish and overshoot it by their own credits (a few cents at most in practice).

## How usage is counted

- Without a subscription, requests run until the month's free 1,000 credits are used, then return `402 payment_required`.
- With a subscription, requests run until the [spend cap](https://webforai.dev/platform/billing#spend-cap).
- Only successful operations are billed, each exactly once — retries on our side never double-count. The dashboard's `Usage` view shows the same records your invoice is built from.

## Managing the subscription

Checkout and the billing portal are both reachable from the [dashboard](https://platform.webforai.dev/dashboard). Cancelling drops you back to the free allowance rather than disabling the account.

## Self-hosting

Billing is optional. Running your own instance without `STRIPE_SECRET_KEY` starts the platform in free-allowance-only mode — everything works, nothing is metered externally, and you pay Cloudflare and your proxy provider directly instead. If you do want metering, `pnpm stripe:setup` creates the meter and the metered price for you. See the [`apps/platform` README](https://github.com/inaridiy/webforai/tree/main/apps/platform#deploying-your-own-instance).