The new Firecrawl /search - Our most accurate Search API for AI agents.
by•
Search is how AI agents ground themselves in the web, but reading full pages for every query burns tokens fast.
We trained a model that returns the excerpts from each /search result that best answer your query, giving your AI agents highly relevant context from every page.
It outperforms processing full pages while using 10x fewer tokens.
On SimpleQA, AI agents using Firecrawl /search now score 94.7%, higher than any other provider. It's live today on every /search call.
Replies
firecrawl has become the default answer every time someone in the agent space asks "how do i give my agent reliable web access." that kind of organic word of mouth in developer communities is hard to manufacture. the "interact with" part beyond just scraping is where it gets interesting for agentic use cases. filling forms, clicking through flows, handling dynamic content that only renders after javascript executes. curious how the rate limiting and retry logic works when a target site starts throttling requests mid-crawl.
The 10x fewer tokens claim is the part that caught my attention. In agent workflows, the expensive failure is often not retrieval itself, but letting the model chew through irrelevant page structure. I’d be curious to see where the relevance model struggles most: tables, long docs, or queries with several equally plausible answers?
94.7% on SimpleQA is impressive. How does the relevance model handle ambiguous queries?
The excerpt approach is a clean tradeoff for factual lookups where the signal is usually in one paragraph. Curious how the model handles research queries where the evidence is spread across 3-4 short paragraphs on the same page, or where two sources directly contradict, does it surface multiple excerpts per result, or one best-match per page? That gap is where we've seen agents get confidently wrong answers from snippet-based search.
does the response expose a per-excerpt relevance score, or just the model's own cutoff? asking because in an agent chain I'd rather set my own threshold than trust a black-box cutoff on which excerpts made it in.
The /search endpoint is the piece I've been waiting for — structured web data straight into the pipeline instead of scraping + cleaning by hand. For SEO/content ops this cuts the "research" step from hours to minutes. What's the rate limit story like at scale?
the 10x fewer tokens number is what'll get this adopted fast. curious about the relevance model's language coverage though - was the 94.7% SimpleQA number an English-only benchmark, and does the excerpt scoring hold up the same way on non-English source pages, or is there a noticeable accuracy gap once the query and the page aren't in English?
a small eval set comparing search quality for agent tasks, not just human search. For example freshness, source diversity, deduping, and whether results are easy to turn into cited answers.
if these numbers are true, holy moly lock your doors and hide. The giants might not like this. But I love that it uses as few tokens as possible. Right now token usage is HUGE. This helps the bubble not pop.