> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pocketutils.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Rate Limits & Credits
> Credit costs per endpoint, plan limits, top-up packs, and the headers every response carries.
PocketUtils meters usage in two independent ways. It's worth keeping them separate, because they answer different questions:
**How much** you may spend before your cycle renews or you top up. Set by your plan.
**How fast** you may spend it, in requests per minute. Also set by your plan.
A request that would exceed either one is rejected before any work happens, and rejected requests never spend a credit.
## Credit cost per endpoint
Endpoints come in two tiers. **Light** endpoints fetch and parse — no browser involved. **Render** endpoints boot headless Chromium, or crawl more than one page.
| Endpoint | Method | Tier | Credits |
| ---------------------------------------------------- | ------ | ------ | ------- |
| [`/screenshot`](/api-reference/screenshot) | POST | Render | 2 |
| [`/url-to-pdf`](/api-reference/url-to-pdf) | POST | Render | 2 |
| [`/broken-links`](/api-reference/broken-links) | POST | Render | 2 |
| [`/broken-links/crawl`](/api-reference/broken-links) | POST | Render | 2 |
| [`/meta-scan`](/api-reference/meta-scan) | GET | Light | 1 |
| [`/ssl-scan`](/api-reference/ssl-scan) | GET | Light | 1 |
| [`/down-check`](/api-reference/down-check) | GET | Light | 1 |
| [`/link-extract`](/api-reference/link-extract) | GET | Light | 1 |
| [`/link-extract/crawl`](/api-reference/link-extract) | POST | Light | 1 |
| [`/xbel-decrypt`](/api-reference/xbel-decrypt) | POST | Light | 1 |
| [`/xbel-to-html`](/api-reference/xbel-to-html) | POST | Light | 1 |
| `/quota` | GET | — | Free |
| `/broken-links/crawl/:jobId` (status) | GET | — | Free |
| `/link-extract/crawl/:jobId` (status) | GET | — | Free |
| `/xbel-decrypt/download/:filename` | GET | — | Free |
| `/xbel-to-html/download/:filename` | GET | — | Free |
Starting a crawl costs credits once, when you start it; polling its status and downloading the resulting file are always free and don't count toward your rate limit either. See [Async Jobs & Polling](/async-jobs).
## Failed requests are never billed
Credits are reserved *before* your request runs and confirmed only after it finishes — so a burst of concurrent requests can never collectively overdraw a balance that each one individually saw as sufficient. If the request then fails (any non-2xx response, or the connection drops mid-render), the reservation is automatically refunded. You are only ever charged for a response that actually succeeds.
## Plans
| Plan | Price | Credits / cycle | Requests / min | API keys | Usage history |
| -------------- | ------- | ----------------- | -------------- | ------------- | ------------- |
| **Free** | \$0 | 100 | 20 | Not available | 30 days |
| **Starter** | \$9/mo | 5,000 | 60 | 2 | 1 year |
| **Growth** | \$29/mo | 20,000 | 300 | 10 | 2 years |
| **Scale** | \$99/mo | 80,000 | 1,200 | Unlimited | 5 years |
| **Enterprise** | Custom | Custom (250,000+) | 5,000 | Unlimited | 5 years |
The Free plan's 100 credits/month is the allowance for a signed-in visitor using the website's tool pages — it does not include API key access. Programmatic access starts on Starter.
### Anonymous (signed-out) allowance
No account at all still gets a small, deliberately limited allowance, tracked per IP:
* **10 credits per day**
* **10 requests per minute**
### Top-up packs
On any paid plan, buy additional credits that don't expire with the billing cycle:
| Pack | Credits | Price |
| ----------- | ------- | ----- |
| `pack_5k` | 5,000 | \$10 |
| `pack_25k` | 25,000 | \$45 |
| `pack_100k` | 100,000 | \$160 |
Cycle credits are spent before top-up credits.
## Response headers
Every `/v1` response carries these headers, whether it succeeds or fails:
Unique ID for this request. Include it when contacting support about a specific call.
Credits this request cost (0 for free endpoints).
Your balance immediately after this request was reserved.
Your plan's requests-per-minute ceiling.
Requests left in the current one-minute window.
Seconds until the window resets.
Light `GET` endpoints also send `Cache-Control: no-store`. Without it, a browser or intermediary could serve a stale scan from cache, or silently revalidate and re-run the whole thing for a `304` — neither is what re-running a scan should do.
## 429 vs. 402: which limit did you hit
| Status | Code | Meaning | What to do |
| ------ | ---------------------- | --------------------------------------------------------- | -------------------------------------------------- |
| `429` | `rate_limited` | Too many requests in the current minute | Back off and retry after `RateLimit-Reset` seconds |
| `429` | `free_quota_exhausted` | Anonymous daily allowance used up | Sign in, or wait for the daily reset |
| `402` | `insufficient_credits` | Signed-in/API-key balance too low for this request's cost | Buy a top-up pack, or wait for cycle renewal |
`402` responses include a `creditsRequired` field so you know exactly how far short you were:
```json theme={null}
{
"success": false,
"error": {
"code": "insufficient_credits",
"message": "Not enough credits for this request. Buy a top-up pack or wait for your cycle to renew.",
"creditsRequired": 2
}
}
```
How each caller type is identified and billed.
The full error code reference.
## Related topics
- [Introduction](/index.md)
- [Get quota](/api-reference/quota.md)
- [Authentication](/authentication.md)
- [Quickstart](/quickstart.md)
- [Errors](/errors.md)