Skip to main content
Everything in Hydrafetch is priced in credits. There are no per-endpoint plans or separate meters to reconcile: a scrape, a crawl, a search, and an extraction all draw down the same balance. Your workspace holds that balance, and any key in the workspace spends from it.

​
Charged only on success

Credits are deducted only when a call actually returns you data. If a request fails — a bad key, an invalid body, an unreachable page, a timeout — it costs nothing.
Because failures are free, you can retry safely. You’re never billed for a call that didn’t produce a result.
Insufficient balance is caught up front. If your workspace can’t cover a call, Hydrafetch rejects it before doing any work and returns an insufficient-credits error — so you’re never charged and no partial work runs.

​
What each call costs

​
1 credit base
One credit for a standard scrape (Markdown, HTML, links, or structured data). Add +1 if you request the summary format and +4 if you request json. See Formats.
​
1 credit
One credit to enumerate a site’s URLs, regardless of how many come back.
​
1 credit + 1 per result scraped
One credit for the query, plus one credit for each result whose page is actually fetched and returned. Turn off result scraping to pay just the single query credit.
​
5 credits per page
Five credits for each page that returns data. Pages that yield nothing aren’t charged.
​
1 credit per page
One credit for each page scraped. The running total is reported on the job’s creditsUsed as it progresses.
A crawl or batch that requests premium per-page formats is priced the same way a scrape is — the base page credit plus any format surcharge — for every page scraped.

​
Every response reports its cost

Each scrape response carries a usage object telling you exactly what it consumed, how much you have left, and whether it came from cache:
​
number
Credits this call consumed. The exact number that left your balance, charged only on success.
​
number
Credits left in your workspace’s balance after this call, so you can track spend without a separate lookup.
​
string
fresh if the page was fetched for this call, or cache if it was served from a recent capture. Cached results are cheaper and faster. See Caching.
For long-running jobs, poll the crawl or batch status and read creditsUsed for the total spent so far.

Next: Caching

Serve recent captures to cut cost and latency.