WebCrawler API logoWebCrawler API
Back to Changelog

Scrape Quality Benchmark

We published a scrape quality benchmark that compares WebCrawlerAPI with Firecrawl, Spider, Jina Reader, Crawl4AI, Context.dev and Crawlbase.

What was measured

  • All tools ran on the same 806 valid URLs from the public firecrawl/scrape-content-dataset-v1 dataset, with default markdown settings.
  • Metrics: end-to-end success, content recall, extraction F1, noise avoidance, error rate and median and p95 latency.
  • Results were collected on 2026-10-12. Per-URL outcomes are available in an interactive explorer.

Where to read it

The full method, metric definitions, limits and scoring code are on the benchmark page. There is also a section to report a problem if you think a result is wrong.