ScrapeGraphAI Pricing for Every Plan and What an Extraction Costs
7 min read
ScrapeGraphAI pricing runs from a free 500 credits to Starter at $20 a month for 10,000 credits, Growth at $100 for 100,000 and Pro at $500 for 750,000. An AI extraction costs 5 credits, so 1,000 pages cost between $3.33 on Pro and $10 on Starter. This page prices three real jobs and shows where a flat record plan comes out ahead.
figures taken from the ScrapeGraphAI pricing page; prices in USD before tax
Robot console
No account needed Running · s Complete · rows
The plans
ScrapeGraphAI plans, prices and credits at a glance
| Plan | Monthly | Credits | Extractions at 5 credits | Per 1,000 extractions | Rate limit | Monitors |
|---|---|---|---|---|---|---|
| Free | $0 | 500, one time | 100 | $0 | 10 a minute | 1 |
| Starter | $20 | 10,000 | 2,000 | $10.00 | 100 a minute | 5 |
| Growth | $100 | 100,000 | 20,000 | $5.00 | 500 a minute | 25 |
| Pro | $500 | 750,000 | 150,000 | $3.33 | 5,000 a minute | 100 |
| Enterprise | priced individually | ad hoc | n/a | n/a | set per account | n/a |
Growth adds basic proxy rotation and Pro adds advanced proxy rotation and priority support. Concurrent crawls climb from 1 on Free to 3, 15 and 50. Credits that run out mid month can be topped up with one-time packs: $5 for 1,000 credits, $40 for 10,000 and $150 for 50,000, which work out to $5, $4 and $3 per 1,000 credits and never expire. Compare that with the plan rates of $2, $1 and about $0.67 per 1,000 and the lesson is plain: top-ups are insurance, not a way to run a job.
The meter
How many credits a ScrapeGraphAI request costs
| Endpoint | What it does | Credits |
|---|---|---|
| Scrape | URL to markdown, HTML, screenshot or branding analysis | 1 markdown, 2 screenshot, 25 branding |
| Extract | Structured fields from a page from a plain language prompt | 5 |
| Search | Search the web and extract from the results in one call | 2 per result, 5 with a prompt |
| Crawl | Walk a whole site and process every page | 2 to start, then the per page cost |
| Monitor | Watch a page and send a webhook on change | Format cost, plus 5 when a change is found |
| Process documents page by page | 1 per page, 25 page default cap |
The part that matters for a recurring job is that Extract is the only endpoint that returns fields, and it is the expensive one. Fetching markdown at 1 credit is cheap, but then your own code or another model call has to turn text into columns. If you are weighing similar AI first APIs, the same five to one gap between a page and a structured page shows up in Firecrawl pricing.
Worked bills
Three real monthly jobs priced on ScrapeGraphAI and WebRobot
| Job | Credits needed | Cheapest ScrapeGraphAI route | WebRobot plan | What ScrapeGraphAI leaves to you |
|---|---|---|---|---|
| A weekly price table of 600 products, four runs a month (2,400 pages) | 12,000 | Starter $20 plus two $5 top-up packs, $30 | Launch, $79 | The schedule, the history table, the sheet |
| 500 product pages every day (15,000 pages) | 75,000 | Growth, $100 | Scale, $249 (Launch caps at 10,000 records) | The schedule, the diff between runs, delivery |
| 3,000 pages every day (90,000 pages) | 450,000 | Pro, $500 | Scale, $249 (100,000 records) | The same, at six times the volume |
Read the first two rows honestly. If a developer already has the glue code, ScrapeGraphAI is the cheaper way to get extracted fields at those volumes, and WebRobot does not win on price. The third row turns on volume: past about 100,000 credits a month the record based Scale plan undercuts the Pro tier. The other variable is people. A job with no developer behind it does not run on any credit API, because the schedule, the retries and the delivery are code someone has to write and fix.
Which plan
Which ScrapeGraphAI plan to buy, or whether to skip it
- Free tests your target. 500 credits is 100 extractions, once.
- Starter ($20) fits a developer prototype of up to 2,000 extractions a month.
- Growth ($100) is the sweet spot, 20,000 extractions at half the unit price of Starter.
- Skip it if nobody on the team writes code, or if the end product is a sheet that must refresh by itself.
The no-code route
WebRobot takes a URL and a plain English list of columns, such as "product, price, stock, rating", and returns rows. The robot follows pagination and detail pages, runs daily on Launch or hourly on Scale, and delivers to Sheets, Excel, Slack or the API. Price monitoring teams start from the price monitoring tool, and the ScrapingAnt pricing and web scraping cost guides price the credit API route in the same format. Developers who do want an endpoint can use the web scraping API.
FAQ
ScrapeGraphAI pricing questions buyers ask
Skip the glue code and get the rows
Describe the columns, check a live run, and get data in Sheets, Excel or the API every day. Plans from $79 a month.