Trace by Context.dev

TRACE is a weekly benchmark that compares web scraping APIs on a fresh sample of about 2,500 pages every Friday. Core capabilities: • Page access rate with confidence intervals • Latency percentiles covering median and slow-tail response times • Document quality checks on returned page content • Cost per page comparison • Failure reason breakdowns • Page-level data for independent verification. Each week the same pages are sent to context.dev, Exa, Firecrawl, and Keenable so results are directly comparable. The benchmark measures whether page data actually loaded, rather than counting blocks, errors, or empty responses, and it reports where each provider falls short. Results and methodology are published openly, with raw data available on GitHub. In the latest sample, context.dev accessed 94.4 percent of pages, ahead of Firecrawl, Keenable, and Exa, while Exa returned the fastest median response. Designed for developers, data teams, and product leads who need to choose a web scraping API based on evidence rather than marketing claims. Use it to evaluate reliability, speed, and cost before committing to a provider, or to track how scraping performance changes over time.