Scraping API PricesDo I need a scraping API or a cloud browser?
For a one-URL-in, one-response-out workload, choose a scraping API, with ScrapingFish providing a published example at $0.002 per successful request; for a stateful click, form, or multi-page workflow, choose a cloud browser, with Kernel’s active headless time priced at exactly $0.06000012 per hour. As of July 26, 2026, sources: https://scrapingfish.com/, https://www.kernel.sh/docs/info/pricing, https://brightdata.com/products/web-unlocker, https://www.browserbase.com/pricing, and https://docs.browserless.io/overview/unit-consumption (retrieved July 26, 2026).
Dated price checked 2026-07-26; cited vendor pages are listed below. This is not a live price check.MethodThe workload decides
A scraping API is the direct fit when each input URL can be submitted independently and the useful output is the returned page, rendered HTML, or structured response. ScrapingFish, for example, publishes a price per successful request and says browser rendering, rotating proxies, and JavaScript execution are available through the same request interface. Bright Data’s Web Unlocker similarly presents a request-and-response product with proxy management, browser rendering, and CAPTCHA handling included. Sources: https://scrapingfish.com/ and https://brightdata.com/pricing/web-unlocker (retrieved July 26, 2026).
A cloud browser is the direct fit when the sequence itself matters: open a page, retain cookies or local state, click, type, wait, navigate, and continue in the same session. Bright Data says Web Unlocker does not support direct interaction with a browser and points interactive use cases to its Browser API. Browserbase prices persistent browser sessions by included and additional browser-hours, while Browserless meters each started 30 seconds of browser time. Sources: https://brightdata.com/products/web-unlocker, https://www.browserbase.com/pricing, and https://docs.browserless.io/overview/unit-consumption (retrieved July 26, 2026).
For 100,000 unrelated public URLs, a request-priced API exposes a countable request unit and avoids paying for a session between independent pages. For one authenticated workflow that visits 20 screens and needs the same cookies, repeatedly recreating state through isolated API calls may not represent the required interaction. These conclusions follow from the products’ documented request and session billing units, not from a claim about vendor speed or reliability. Sources: https://scrapingfish.com/, https://brightdata.com/pricing/web-unlocker, https://www.browserbase.com/pricing, and https://docs.browserless.io/overview/unit-consumption (retrieved July 26, 2026).
The price comparison needs more than a URL count. For an API, record successful billable responses, credit multipliers, rendered-page use, and target-specific charges. For a browser, record active session time, rounding, transferred proxy data, concurrency, and extras such as CAPTCHA attempts. Browserless explicitly adds proxy units per megabyte and CAPTCHA units per attempt to its time units; Browserbase lists plan allowances, overage hours, concurrency, and proxy charges separately. Sources: https://docs.browserless.io/overview/unit-consumption and https://www.browserbase.com/pricing (retrieved July 26, 2026).
Some products cover both patterns, so the product label alone is not decisive. Select the interface and meter that match the job: independent responses favor request accounting, while stateful interaction favors session accounting. A defensible break-even calculation requires the actual request count, rendered share, session duration, bandwidth, and billing term. Sources: https://brightdata.com/products/web-unlocker, https://scrapingfish.com/, https://www.browserbase.com/pricing, and https://docs.browserless.io/overview/unit-consumption (retrieved July 26, 2026).