Perplexity Search
perplexity_searchGet ranked search results from Perplexity’s continuously refreshed index with advanced filtering and customization options.
How it works ↓Pricing
Per call
$0.02
Model
flat
Pay only for what you use. No subscriptions.
Inputs
last_updated_before_filter
stringcountry
stringmax_tokens
integersearch_language_filter
arrayquery *
stringmax_tokens_per_page
integersearch_after_date
stringmax_results
integersearch_domain_filter
arraysearch_recency_filter
stringsearch_before_date
stringlast_updated_after_filter
stringInput Parameters
Cost per run
Execution cost$0.02
About Perplexity Search
Perplexity Search on xpay — agent-grade web search, one MCP call
Perplexity's Search API is built for AI agents. Unlike Google or Bing-style result links, it returns answer-grade snippets with inline citations — the same retrieval layer that powers Perplexity's Sonar chat models, exposed directly so your agent can run search as a discrete step.
xpay wraps Perplexity Search as the perplexity_search MCP tool. Pass a query, get back ranked results with content snippets and source URLs, billed per call.
What you get back
{
"results": [
{
"title": "...",
"url": "https://...",
"snippet": "Answer-grade extract relevant to the query.",
"published_date": "2026-04-22"
}
]
}
Snippets are extractive and answer-shaped, not link-list teasers. That makes Perplexity Search materially better than scraping a SERP when your agent needs to reason over the result content directly.
When Perplexity Search beats Tavily / Exa / Linkup
- Citation requirements. Each result has a verifiable source URL.
- Recency. Perplexity prioritizes fresh content; good for news, regulatory, market-event queries.
- Mixed open-web coverage. Reaches forums, news, docs, and long-tail sources Google indexes but most agent search APIs filter out.
When to choose alternatives
- For paper/research search, Semantic Scholar or arXiv tools on xpay are better.
- For clean Markdown extraction of a known URL, Jina Reader or Firecrawl Scrape are cheaper.
- For vector-search over your private docs, run your own retrieval.
Pricing
Roughly $0.005–$0.02 per call on xpay. Fixed per-call, regardless of result count. $5 free credit on signup.
How it pairs with chat completions
Common agent pattern: perplexity_search to gather sources, then a frontier LLM (anthropic_claude, openai_gpt) to reason over the snippets. Or use perplexity_chat_completions for the all-in-one call where Sonar handles search + answer in a single round-trip.
Frequently Asked Questions
It is Perplexity's standalone web-search endpoint — the same retrieval layer that powers Sonar chat models, exposed directly. You pass a query and get back ranked results with answer-grade snippets and source URLs.
Search returns ranked results + snippets only — no LLM-generated answer. Sonar chat completions runs the search and returns an LLM-synthesized answer plus citations. Use Search when you want raw retrieval (and to feed your own LLM), use Sonar chat for one-call answer + sources.
Approximately $0.005–$0.02 per call on xpay, fixed regardless of result count. New accounts get $5 free credit.
Yes when available. Perplexity surfaces published_date for news and dated content; for static pages the field may be absent.
Tavily is cheaper and good for general agent search with content extraction; Exa is best for neural / "find similar" semantic search. Perplexity Search is best when citation quality, recency, and answer-shaped snippets matter most.
Yes — that is the point of running it through xpay. You authenticate with your xpay key only; xpay holds the upstream Perplexity credential and bills you per call.