$ check 499-risk
499 Timeout Risk Checker
When ChatGPT, Claude or Perplexity needs your page mid-answer, a live fetcher requests it right then, and it doesn't wait around. Too slow and the fetcher abandons the request, your server logs a 499, and the answer gets built from someone else's page. This times any URL the way those fetchers experience it, cold cache included. The background here.
New to this?
How does this work?
It fetches your URL 3 times, once each as ChatGPT-User, Claude-User and Perplexity-User, and rates the median time to first byte against the sub-500ms target from the 499 research. A 4th request carries a cache-buster so the CDN has to go back to origin, which is what most of a crawler's fetches get, since crawlers pull the deep pages nobody has cached. It also walks the redirect chain and weighs the raw HTML.
Everything is measured from Cloudflare's edge in one short session, so read the numbers as directional. The place that knows for certain is your own access log, where an abandoned fetch shows up as a literal 499 line against the fetcher's user agent.
Speed is only half the door. Check what your firewall and robots.txt are doing to the same crawlers with the AI Crawler Access Checker, or see whether AI agents can discover what your site offers with the Agentic Resource Discovery Checker.