Web search and read
Search the live web and read any page as clean text.
Tools2
web__searchfrom 14 credits / searchSearch the live web and return ranked results with source URLs. When to use: you need current facts, sources, or links. Returns ranked results with title/url and (by default) a short text snippet per result; pass include_highlights for query-relevant sentences. This tool does NOT synthesize an answer — read the returned materials yourself. To read one page in full, use web__scrape. Billed per search, covering up to 10 results with their snippets and highlights, and again per result above 10 — counted on the num_results you ask for, not on how many come back. A failed request is not charged.Search the live web and return ranked results with source URLs. When to use: you need current facts, sources, or links. Returns ranked results with title/url and (by default) a short text snippet per result; pass include_highlights for query-relevant sentences. This tool does NOT synthesize an answer — read the returned materials yourself. To read one page in full, use web__scrape. Billed per search, covering up to 10 results with their snippets and highlights, and again per result above 10 — counted on the num_results you ask for, not on how many come back. A failed request is not charged.
- Search, up to 10 results14 credits / search
- Each result above 10+2 credits / result
web__scrapefrom 5 credits / pageFetch one web page and return its clean text (single-page scrape). When to use: you have a specific HTTPS URL and want its readable content. Prefers a live fetch: a cached copy is returned only while the site's own Cache-Control still marks it fresh, or for a few domains kept cached long-term. Returns page text (capped; textTruncated flags a clip) plus optional highlights. For discovery across many pages, use web__search first. No multi-page crawl is offered. Billed per page scraped, and again per page with include_highlights. A failed request is not charged.Fetch one web page and return its clean text (single-page scrape). When to use: you have a specific HTTPS URL and want its readable content. Prefers a live fetch: a cached copy is returned only while the site's own Cache-Control still marks it fresh, or for a few domains kept cached long-term. Returns page text (capped; textTruncated flags a clip) plus optional highlights. For discovery across many pages, use web__search first. No multi-page crawl is offered. Billed per page scraped, and again per page with include_highlights. A failed request is not charged.
- Read a page5 credits / page
- Highlights on a page+5 credits / page
