Web scraping API benchmark
Success rates and latency over the rolling last 30 days.
| Vendor | Success rate | Anti-bot challenge (standard / premium) | Median seconds | P90 seconds | Runs | Last run (UTC) |
|---|---|---|---|---|---|---|
| ScraperAPI | 83.3% | 0.0% / n/a (not on free plan) | 5.87 | 55.74 | 26 | 2026-10-18T19:12:31.239338+00:00 |
| ScrapingBee | 83.3% | 0.0% / n/a (no valid runs yet) | 1.95 | 10.51 | 26 | 2026-10-18T19:09:36.955818+00:00 |
Standard mode = the default request; premium = the vendor's stealth/ultra-premium option where our plan allows it.
Methodology: daily automated tests on public test pages built for scraper testing (toscrape.com, scrapingcourse.com) and robots-allowed pages, using the same targets for every vendor on free-trial plans. Success requires a successful response and expected visible page content. Challenge pages run in standard and premium modes; other targets run in standard mode only. Premium enables JavaScript rendering with ScrapingBee stealth proxies or ScraperAPI ultra-premium proxies; other vendors use their existing JavaScript rendering. Success rates exclude unavailable premium options and legacy 60-second premium timeouts. Latency includes all attempts, including failures and both challenge modes; P90 uses nearest rank. “—” means no observations for that metric.