Cloudflare Bypass Actors: Real 30-Day Failure Rates

Last updated: November 2, 2026 · Data snapshot: November 2, 2026 (trailing 30-day window) · Weekly refresh

Table 1 — Dedicated bypass / unblocker Actors

Actors whose product IS bypassing anti-bot (name or title says bypass/unblock/Cloudflare/captcha), with at least 500 runs in 30 days. Manually reviewed 2026-10-23: vertical scrapers whose titles only advertise "Bypass NNN cap/limit" (site result caps, not anti-bot) are excluded here and appear in Table 2 instead. Thresholds: tables display every matched Actor above 500 runs (display threshold); rankings and medians in this report only include Actors with ≥1,000 runs (ranking threshold). Failure rates across different targets are an ecosystem signal, not a controlled head-to-head benchmark.

Table 2 — Scrapers that advertise anti-bot capability (worst 15 by failure rate)

Matched on anti-bot keywords in name/title/description (bypass, unblock, Cloudflare, captcha, anti-bot, challenge, WAF), ≥1,000 runs/30d. Note: for some vertical scrapers here, "Bypass" in the title refers to the target site’s result cap, not anti-bot protection.

Table 3 — Same battlefield, no bypass branding: general-purpose web crawlers (worst 10)

Popular generic crawlers (cheerio/web/website-content scrapers, ≥500 runs/30d) for baseline comparison.

Key findings

• Dedicated bypass Actors span 0.00% to 3.26% measured 30-day failure rates — memo23/apify-linkedin-search-results-scraper ran 976 times with 0 failures, while scrapeunblocker/scrapeunblocker failed 3.26% of 64,061 runs.
• The highest-failure Actor that advertises anti-bot capability is xtracto/ft-scraper at 22.89% of 4,137 runs — more than 514x the general-crawler median.
• Counter-intuitively, the worst performer in the entire comparison is not a bypass Actor: apify/ai-web-scraper failed 8.35% of 10,293 runs. That is one LLM-driven crawler in one 30-day window — enough to question the "AI crawlers just work" assumption, but not evidence about AI crawling as a category. Meanwhile the most-used general crawler (radeance/similarweb-scraper) fails only 0.00%.
• Bypass cohort median (0.09%) is ~2x the general-crawler median (0.04%): hard targets fail more, but on managed infrastructure the absolute numbers stay low.
• Cheap ≠ unreliable: several sub-$1/1k scrapers in Table 2 run under 1% failure. In this snapshot price did not predict failure; whether maintenance quality does is a working hypothesis our weekly snapshots will test.

How we measure

Conclusion