ScrapeGraphAI Pricing: Plans and Credits Guide

TL;DR

ScrapeGraphAI pricing is credit based. As of June 16, 2026, the Free plan includes 500 credits, Starter includes 10,000 credits for $20/month, Growth includes 100,000 credits for $100/month, and Pro includes 750,000 credits for $500/month. Use the pricing page to choose a plan and the price calculator to estimate a real workload.

scrape starts at 1 credit for markdown, html, links, images, or summary.
extract costs 5 credits per call before optional stealth.
search costs 2 credits per result without extraction or 5 credits per result with a prompt.
crawl costs 2 credits to start, plus the per-page scrape formats.
monitor bills each tick at the selected scrape format cost, plus 5 credits when a change is detected.

The old "one page always equals ten credits" rule is not the current model. Credits now depend on the service, output formats, page count, search result count, and whether stealth is enabled.

This guide is written for planning, not procurement. Treat the public pricing page as the checkout source of truth and use this article to understand why two scraping jobs with the same number of URLs can consume different credits.

Current Plans

Plan	Monthly price	Yearly price	Included credits	Limits
Free	$0	$0	500 one-time credits	10 requests/min, 1 monitor, 1 concurrent crawl
Starter	$20/month	$204/year	10,000 credits/month	100 requests/min, 5 monitors, 3 concurrent crawls
Growth	$100/month	$1,020/year	100,000 credits/month	500 requests/min, 25 monitors, 15 concurrent crawls
Pro	$500/month	$5,100/year	750,000 credits/month	5,000 requests/min, 100 monitors, 50 concurrent crawls

Ready to scrape?

Start for free

Credit packs are available when you need more credits without changing plans: 1,000 credits for $5, 10,000 credits for $40, and 50,000 credits for $150.

For checkout details, annual discounts, and the live plan table, use ScrapeGraphAI pricing. For workload math, use the standalone price calculator.

How Credits Are Charged

ScrapeGraphAI V2 separates web data work into five services. That matters because each service has a different cost shape.

Service	Best for	Credit rule
`scrape`	Known URL, page formats, screenshots, links, Markdown, HTML, or JSON inside one fetch	Sum the selected format costs
`extract`	Structured JSON from a URL, HTML, or Markdown	5 credits per call
`search`	Web discovery plus optional extraction	2 credits per result, or 5 per result with a prompt
`crawl`	Multi-page site traversal	2 credits to start, plus per-page `scrape` cost
`monitor`	Scheduled page checks and diffs	Per-tick `scrape` cost, plus 5 credits when changed

For scrape, format costs are additive:

Format	Credits
`markdown`	1
`html`	1
`links`	1
`images`	1
`summary`	1
`json`	5
`screenshot`	2
`branding`	25

If one request asks for Markdown, links, and a screenshot, the base cost is 1 + 1 + 2 = 4 credits. If it asks for Markdown and JSON extraction in one scrape call, the base cost is 1 + 5 = 6 credits.

fetchConfig.mode can be auto, fast, or js. Render mode does not add credits. fetchConfig.stealth adds 5 credits because it uses the more expensive anti-bot path.

Cost Examples

These examples use the current credit rules. The price calculator is still the best place to model exact volumes because it applies the same shared pricing logic used by the app.

Convert Known Pages To Markdown

For known URLs, call sgai.scrape with markdown when you need clean text for a database, RAG pipeline, or agent context.

Markdown workload	Formula	Credits
500 pages to Markdown	500 x 1	500
10,000 pages to Markdown	10,000 x 1	10,000
10,000 pages to Markdown with stealth	10,000 x (1 + 5)	60,000

Starter can cover roughly 10,000 simple Markdown pages per month before add-ons. Growth can cover roughly 100,000 simple Markdown pages.

Extract Structured Product Data

Choose sgai.extract when the output needs typed fields like name, price, availability, and productUrl.

Extraction workload	Formula	Credits
1,000 product pages	1,000 x 5	5,000
10,000 product pages	10,000 x 5	50,000
10,000 product pages with stealth	10,000 x (5 + 5)	100,000

This is why the right service matters. If you only need Markdown, use scrape and avoid the LLM extraction cost. If you need reliable business objects, extract is the right path.

Search And Extract Market Data

Reach for sgai.search when the input is a query instead of a fixed URL list.

Search workload	Formula	Credits
100 searches, 5 results each, no prompt	100 x 5 x 2	1,000
100 searches, 5 results each, with prompt	100 x 5 x 5	2,500
1,000 searches, 10 results each, with prompt	1,000 x 10 x 5	50,000

Search is useful for market research, competitor discovery, vendor monitoring, and lead enrichment when the source pages are not known upfront. If you already have the URLs, scrape or extract is cheaper and more deterministic.

Crawl A Website

Start a sgai.crawl.start job when one URL needs to expand into many linked pages.

Crawl workload	Formula	Credits
Crawl 100 pages to Markdown	2 startup + 100 x 1	102
Crawl 1,000 pages to Markdown and links	2 startup + 1,000 x (1 + 1)	2,002
Crawl 1,000 pages to JSON	2 startup + 1,000 x 5	5,002

Crawl is async. Start the job with sgai.crawl.start, poll it with sgai.crawl.get, and fetch page results from the crawl pages endpoint or SDK helpers.

Monitor A Page

Create recurring checks with sgai.monitor.create. Each tick costs the selected format cost. If a tick detects a change, add 5 credits.

Monitor workload	Formula	Credits
One hourly Markdown monitor for 30 days, no changes	720 x 1	720
One hourly Markdown monitor for 30 days, 20 changed ticks	720 x 1 + 20 x 5	820
Ten hourly Markdown monitors for 30 days, no changes	10 x 720 x 1	7,200

For price tracking, inventory alerts, job listing watches, or brand mention checks, monitor is often cheaper and cleaner than running your own scheduler.

Which Plan Fits Which Workload?

Use the Free plan to test extraction quality and integration paths. It is not a production volume plan, but 500 credits is enough to validate prompts, schemas, fetch settings, and response handling.

Use Starter when the workflow is small and predictable: a few thousand extract calls, about ten thousand simple Markdown pages, or a handful of monitors.

Use Growth when the workload is a real data pipeline: frequent market research searches, larger URL lists, a set of scheduled monitors, or multi-page crawls. Growth also includes higher rate limits and more concurrent crawl and monitor capacity.

Use Pro when credits are not the only constraint. Pro is for teams that need much higher request rates, more monitors, more concurrent crawls, advanced proxy rotation, and priority support.

Enterprise is for custom volume, support, security, or procurement requirements. If your team needs custom limits or a private commercial agreement, the public plan table is only the starting point.

How To Keep Costs Predictable

Choose the cheapest service that returns the output you actually need. Use scrape for page formats, extract for typed JSON, search for discovery, crawl for multi-page traversal, and monitor for scheduled checks.

Request multiple scrape formats in one call when the formats belong to the same page. Markdown plus links plus screenshot should be one sgai.scrape call, not three separate fetches.

Avoid enabling stealth for every request. Start with default fetching, inspect failures, then enable stealth only for targets that need it. The extra 5 credits per request or page becomes meaningful at scale.

Keep prompts specific. A precise extract prompt with a schema wastes fewer retries than a vague prompt that returns ambiguous fields.

Model the workload before committing to a plan. The price calculator exists for this exact reason, and the public pricing page shows the current plan limits.

If the first estimate is close to a plan limit, test a representative sample before buying for the whole volume. A page that needs JavaScript rendering, scrolling, screenshots, JSON extraction, or stealth can have a different credit profile than a static page converted to Markdown.