> ## Documentation Index
> Fetch the complete documentation index at: https://docs.spidra.io/llms.txt
> Use this file to discover all available pages before exploring further.
# How Credits Work
> What a Spidra credit pays for: scrapes, crawls, batches, search, AI extraction and CAPTCHA solving, plus how proxy bandwidth is metered, with worked examples.
Spidra charges in credits. One balance covers [scraping](/api-reference/scraping/scrape), [crawling](/api-reference/crawling/crawl) and [search](/api-reference/search/search), so you never buy them separately. This page explains what each action costs so you can predict a bill before you run a job.
Proxy usage is metered separately in bandwidth, not credits. That is covered [further down](#proxy-bandwidth).
## What costs credits
| What you do | Cost |
| - | - |
| Scrape a page or document | 1 credit per URL |
| Crawl a site | 1 credit per page crawled successfully |
| Use AI extraction (a `prompt`, `schema` or `transformInstruction`) | Extra credits based on the tokens the AI uses |
| Have a CAPTCHA solved | 5 credits per solve |
| Search | 1 credit per 10 results, per source |
| Fetch page content for search results | 1 credit per page fetched |
| Run browser actions | Free |
| Run a job that fails | Free |
## Scraping
A scrape costs **1 credit for each URL that is processed successfully**. Send three URLs in one request and all three succeed, and that is 3 credits. The size of the page does not change the price.
If you ask for AI extraction, the tokens it uses are added on top. Extraction runs when you set a `prompt` or a `schema`, or set `output` to `"json"`. About 4 characters of page content count as 1 token, and roughly 3,300 input tokens or 400 output tokens come to 1 credit. Token credits are added up for the whole job and rounded up once, so a short page with a short answer usually adds a single credit.
If Spidra solves a CAPTCHA on the way, that adds 5 credits per solve. A page with no challenge costs nothing extra. See [CAPTCHA solver](/features/captcha-solver).
[Documents](/api-reference/scraping/scrape#document-urls-pdf-word-excel-powerpoint) such as PDFs, Word and Excel files are charged like pages. A scanned document that needs AI to transcribe it also uses tokens.
| Example | Credits |
| - | -: |
| 1 URL, no AI | 1 |
| 3 URLs in one request, no AI | 3 |
| 1 URL with a short extraction prompt on a normal page | 2 |
| 1 URL where a CAPTCHA was solved, no AI | 6 |
| A long PDF with an extraction prompt | 1 plus the tokens it takes, often 5 or more |
## Batch scraping
Each URL in a [batch](/features/batch-scraping) is charged as its own scrape. Failed URLs cost nothing and the rest are charged normally, so a batch of 50 where 45 succeed costs for 45.
## Crawling
A [crawl](/features/crawling) costs **1 credit for every page it crawls successfully**, plus tokens if you give it a `transformInstruction` or a `schema`. Token credits are added up across all pages and rounded up once.
| Example | Credits |
| - | -: |
| 10 pages, no extraction | 10 |
| 10 pages with a `transformInstruction` | 10 plus the tokens used |
Re-running extraction on pages you already crawled is cheaper, because it only charges for tokens and not for the pages again. See [Extract from a crawl](/api-reference/crawling/crawl-extract).
## Search
[Search](/features/search) costs **1 credit per 10 results, per source, rounded up**. A source that returns nothing costs nothing.
| Example | Credits |
| - | -: |
| Web search, 10 results | 1 |
| Web search, 20 results | 2 |
| Web, news, images and videos with 5 results each | 4 |
| A search that returns no results | 0 |
If you also ask Search to fetch the content of each result with `scrapeOptions`, every page fetched adds 1 credit. Fetching content for 5 results from a 10 result search costs 1 for the search plus 5 for the pages, so 6 in total.
## What is always free
* **Browser actions.** Clicks, typing, scrolling, waiting and `forEach` loops never cost credits. See [Actions](/features/actions).
* **Failed jobs.** If a scrape, batch item or crawl cannot deliver anything, you are not charged. That includes a page that stays behind a bot-protection challenge after Spidra has tried to get past it.
* **Reading things.** Checking a job's status, listing [logs](/features/logs) and reading your usage cost nothing.
## Proxy bandwidth
[Stealth Mode](/features/stealth-mode) routes a request through a residential proxy. It does not cost extra credits. Instead it uses your **proxy bandwidth**, a separate allowance measured in MB of data downloaded.
A typical page uses around 2 MB. Pages heavy with images, video or ads can use 20 MB or more, because the browser downloads everything on the page and not just the text.
Each plan includes an amount (50 MB on Free up to 5 GB on Pro, see [Plans and Pricing](/billing/plans)). If you need more, buy a [bandwidth top-up](/billing/top-ups#proxy-bandwidth-top-ups).
If you ask for Stealth Mode on a scrape (`useProxy: true`) and you have no bandwidth left, the scrape still runs, but without the proxy. It uses a direct connection, so it may be blocked on sites that need a proxy. Check your balance before large proxy jobs.
## Credits are held while a job runs
When you submit a job, Spidra holds the most it could cost and returns whatever is not needed when the job finishes. While a job is running, the held credits already count toward your balance. If your balance cannot cover the hold, the request is rejected with a `403` before anything runs.
## When you run out
Once your plan credits are used up, Spidra uses any [top-up credits](/billing/top-ups) you have. With none left, new jobs fail with a `403` and the message "You have exceeded your monthly credit limit". You have three options:
* On a paid plan, buy a [top-up](/billing/top-ups) and carry on straight away.
* [Upgrade your plan](/billing/managing-subscription#upgrading-your-plan).
* On a paid plan, wait for your credits to reset at the next billing cycle. The Free plan's credits do not reset.
The full list of status codes is in [Errors](/api-reference/errors).
## Check your balance
* **Dashboard:** the [Usage page](https://app.spidra.io/dashboard/usage) shows your credits and proxy bandwidth, with a daily chart. There is also a Credit Usage meter at the bottom of the sidebar.
* **API:** [`GET /account`](/api-reference/account/account) returns what you have left, and [`GET /usage-stats`](/api-reference/account/usage-stats) shows usage over time.
* **Per job:** every scrape and search in your [logs](/features/logs) lists the exact credits it used.
What each plan includes
Add credits or bandwidth on top of your plan
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.