Credit billing

Credits pay for operations with a per-request or per-upstream-page cost. They are tracked separately from bandwidth.

Only the two surfaces below are credit-billed. Scrape, fetch, screenshot, sitemap, and monitor-run requests are billed on bandwidth instead — see Bandwidth billing for what those endpoints charge.

Search

Search costs one credit for every upstream results page requested:

pages = min(ceil(limit / 10), 10)
credits = pages

For example, a limit of 10 costs one credit, a limit of 11 costs two credits, and limits up to 100 are capped at ten upstream pages and ten credits. The charge follows the requested limit even if an upstream page returns fewer results.

Schema extraction

Schema extraction is dual-axis billing. Each request has a fixed 10-credit extraction charge and also consumes bandwidth for its source-page fetch. ByteKit pre-charges an estimated 100 KiB for that fetch, then reconciles the bandwidth charge to the actual upstream wire bytes.

If the request fails before the source page is fetched, ByteKit compensates the credit and bandwidth pre-charges. Failures after the fetch may retain the source-page bandwidth charge because that upstream work was completed.

Your effective credit budget includes plan credits and purchased credits. Credits do not roll over — they reset every billing period.