Parallel tools

Return ranked companies matching a plain-language description, each with a URL and summary.

Return ranked people matching a plain-language description, each with a profile URL and summary.

Return ranked pages with query-focused excerpts in basic or advanced mode.

Return ranked pages with query-focused excerpts in turbo or fast mode.

Answer up to 100 typed questions about a subject from live web evidence.

Reads public web pages you already have and returns the parts relevant to an objective or keyword queries. Returns each page's URL, title, publish date and markdown excerpts, optionally the whole page as markdown, and a separate list of URLs that could not be read with the error type and HTTP status. Supports cached or live fetch with a freshness bound, excerpt and full-content size limits, and a per-URL live-fetch wait of up to 30 s. Only pages read successfully are billed. Suited for reading search results, articles, docs and product pages, and feeding page text to an LLM.

Searches the web for pictures matching 1-5 keyword queries and an optional natural-language description of what should be in the image. Returns up to 20 results, each with the direct image URL, the page it appears on, a title or caption, and pixel width and height. Supports fast (~1 s) and advanced (~3 s, higher quality) modes and country targeting. Suited for illustrating answers, product and place photos, landmarks, and collecting visual references for a topic.