LLM Reader: Web Page to Markdown API for RAG
Pricing
from $3.50 / 1,000 web page converteds
Convert one public web page into clean main-content Markdown for RAG, LLMs and AI agents. Extract title, author, source URL and estimated tokens using HTTP fetching with optional JavaScript rendering. Structured JSON output; no whole-site crawling.
Absolute public http:// or https:// URL. Credentials, private/special IP ranges, local hostnames, and ports other than 80/443 are rejected.
When true, launch Playwright immediately. When false, use lightweight HTTP first and fall back to Playwright only if browser_fallback is enabled and needed.
Direct Actor option: when render_js is false, allow Playwright only if lightweight extraction fails or is too short. Set false for a strict low-cost, fetch-only run.
Lightweight results shorter than this trigger browser fallback. This threshold does not pad or fabricate content.
Opt in to protocol/shape-filtered but untrusted HTTP(S) and mailto references. Relative links are resolved; public-looking hostnames are not DNS-validated, so downstream consumers must validate before fetching.
Opt in to protocol/shape-filtered but untrusted image references. Image binaries are never downloaded by the browser renderer; downstream consumers must validate references before fetching.
Per-request timeout. The global extraction deadline can shorten it, especially before a browser fallback.
Maximum Playwright navigation budget. The global extraction deadline always takes precedence.
Application deadline shared by HTTP, browser, and parsing phases. It starts after the Actor process begins and excludes container allocation and cold-start time; surrounding callers must configure a larger end-to-end deadline.
