> ## Documentation Index
> Fetch the complete documentation index at: https://docs.zapfetch.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Rate limits

> Per-plan request limits, 429 responses, and recommended backoff.

ZapFetch applies per-API-key rate limits to protect the platform and your
own credit balance from runaway loops. Limits are enforced globally across
all ZapFetch endpoints.

## Per-plan limits

> These numbers are indicative. The authoritative values live in the
> [Console](https://console.zapfetch.com/dashboard/billing) under your current plan.

| Plan    | `/scrape`, `/search`, `/map`, `/extract` (req/sec) | `/crawl` (req/sec) | Concurrent crawls |
| ------- | :------------------------------------------------: | :----------------: | :---------------: |
| Free    |                          5                         |          2         |         2         |
| Starter |                         50                         |         10         |         10        |
| Pro     |                         200                        |         30         |         20        |

`/v1/crawl` has a tighter per-second budget because each call can dispatch
hundreds of background page fetches. Concurrent crawls cap the number of
`/v1/crawl` jobs *in-flight* for a single key; completed or cancelled
crawls don't count.

## 429 response shape

When you exceed a limit ZapFetch returns `429 Too Many Requests` with a
`Retry-After` header (seconds):

```http theme={null}
HTTP/1.1 429 Too Many Requests
Retry-After: 12
Content-Type: application/json

{
  "success": false,
  "error": "rate_limited",
  "message": "You have exceeded 50 requests/second for the Starter plan."
}
```

## Recommended client-side behavior

* Honor `Retry-After` literally — don't retry before the header says you can.
* On repeated `429`s, fall back to exponential backoff with jitter:
  `sleep(min(cap, base * 2^attempt) + random_jitter)`.
* Batch work where possible: `/v1/crawl` accepts a `limit` and `maxDepth`
  so you do not have to orchestrate per-page `/v1/scrape` calls yourself.
* Pool connections per-process — a burst of "one-shot" calls hits rate
  limits faster than a steady stream with keep-alive.

## Requesting higher limits

Pro customers with bursty workloads can request a higher per-minute ceiling
by emailing `support@zapfetch.com`; we review case-by-case. Rate limits
never exceed what your credit balance can actually support — this is a
safety net, not a bottleneck for normal use.