Perplexity Search API
Call Perplexity's standalone Search API through LLM Gateway at /v1/search. It returns ranked web results with extracted page snippets instead of a model-written answer, billed per request, with failed requests free.
Agents and retrieval pipelines often need the raw search results, not a model's
summary of them, and until now the only Perplexity product on the gateway was
Sonar, which always answers in prose. The Perplexity Search API is now
available at /v1/search, returning ranked web results with extracted page
content that you feed into your own prompts and tools.
A drop-in for Perplexity Search
The request and response match Perplexity's Search API, so existing code only
needs the LLM Gateway base URL and API key. Every field except model is
forwarded to Perplexity unchanged, including domain, language, country, recency
and date filters.
1curl https://api.llmgateway.io/v1/search \2 -H "Authorization: Bearer $LLM_GATEWAY_API_KEY" \3 -H "Content-Type: application/json" \4 -d '{5 "query": "latest developments in open-source LLMs",6 "max_results": 5,7 "search_recency_filter": "week"8 }'1curl https://api.llmgateway.io/v1/search \2 -H "Authorization: Bearer $LLM_GATEWAY_API_KEY" \3 -H "Content-Type: application/json" \4 -d '{5 "query": "latest developments in open-source LLMs",6 "max_results": 5,7 "search_recency_filter": "week"8 }'Each result carries a title, url, snippet, and publication and last-updated
dates. Pass query as an array of up to five related queries to search them
independently in one request.
Two models, billed per search
model is optional. When you leave it out, Perplexity's search_type picks the
model:
search_type | Model | Price |
|---|---|---|
web (or omitted) | perplexity/perplexity-search | $5 per 1,000 searches |
fast | perplexity/perplexity-search-fast | $1 per 1,000 searches |
- No token charges. Search is billed per successful request, however much page content comes back.
- Multi-query requests bill once, matching Perplexity's own billing.
- Failed requests are free. Upstream errors are logged at no cost.
search_type: "people"is not supported yet and is rejected with a400.
Requests appear in your activity log and usage analytics like any other endpoint, and respect provider keys, IAM rules and data retention settings.