Perplexity News Citations as Scrapeable Data
Perplexity News search results are useful raw material for news monitoring, competitor tracking, and research pipelines, but the pages are not designed to be pulled at scale.
A hosted scraping API takes a Perplexity News search URL and returns a rendered, parsed response from one request, with proxy rotation, browser rendering, and anti-bot handling handled on the provider side. This page explains what the API returns, how to shape the output into JSON, what each request option costs in credits, and where the practical limits sit.
What a Perplexity News search page contains
Each Perplexity News search result carries a small set of fields that are worth extracting: the headline, the source publication, a snippet, the publish timestamp, the AI-cited URLs, and the article URL. Those six fields are enough to build a news feed, a citation graph, or a competitor watchlist without storing the full page. The API returns the rendered page; you decide which fields to keep by mapping CSS selectors to field names with extract_rules, or by describing the fields in natural language with ai_extract_rules. The response comes back as parsed JSON, so downstream steps can load it directly into a database, spreadsheet, or LLM prompt.
Three workflows the API supports
The most common uses fall into three groups. First, trending-topic extraction: pull the current Perplexity News topics and compare them against your own market map to spot shifts early. Second, article-metadata collection: keep headline, publication, snippet, timestamp, and citation URLs for a news-research pipeline. Third, author social-link collection: capture the social profiles attached to articles so you can analyze who is driving a story. Each workflow uses the same request shape; only the extract_rules or ai_extract_rules differ.
How a request works
You need an API key and a Perplexity News search URL. Send both to the endpoint; the service rotates proxies, spoofs browser fingerprints, and renders JavaScript before returning a clean response. If the page needs more than the default, you can add render_js for JavaScript rendering, stealth_proxy for tougher anti-bot pages, country_code for geo-targeting, and ai_extract_rules for natural-language extraction. The dashboard request builder generates a working snippet per language, so you can test the call before dropping it into your codebase.
Field map for a Perplexity News result
Use this checklist when writing your extract_rules or ai_extract_rules. Each row is a field the source page is expected to expose; the selector column is intentionally left for you to fill in against the live page, because class names change.
- Headline — the article title shown in the result card.
- Source publication — the outlet credited with the article.
- Snippet — the short summary text under the headline.
- Publish timestamp — when the article was published.
- AI-cited URLs — the citation links Perplexity attaches to the result.
- Article URL — the canonical link to the full article.
If a field is missing from a given result, return null rather than dropping the row; downstream joins are easier when the schema is stable.
Worked example: shaping one result into JSON
Suppose a Perplexity News search returns a single article about a funding round. Using the field list above, a parsed JSON object would look like this in shape (values are illustrative, not captured from a live call):
{ "headline": "Example startup raises a Series A", "source_publication": "Example News", "snippet": "The company said the round will fund hiring.", "publish_timestamp": "2026-01-01T00:00:00Z", "ai_cited_urls": ["https://example.com/article"], "article_url": "https://example.com/article" }
To produce this, map each field name to a CSS selector in extract_rules, or describe the same six fields in ai_extract_rules. The response is parsed JSON, so you can write it straight to storage. Keep the field names stable across runs so your schema does not drift.
Credit costs per request
Credit consumption depends on which options you enable. The base request starts at 1 credit. JavaScript rendering adds 5 credits per successful request. Premium proxies cost 10 credits without JS rendering and 25 with it. Stealth proxy costs 75 credits. The table below summarizes the combinations described in the pricing tooltips and FAQ; treat the numbers as the published reference and confirm current pricing before committing to a plan.
Concurrency, not per-target rate limits
The API applies concurrency limits per plan rather than a per-target rate limit. If the target site throttles aggressive patterns, space out requests or use multiple sessions. Concurrency ranges from 25 on the entry plan up to 900 on the largest enterprise tier; the exact numbers are in the plan table. There is no separate per-domain quota to configure, so the practical ceiling is your plan's concurrent request count.
Languages and SDKs
The service publishes SDKs for Python, Node.js, Go, PHP, Ruby, and Java, and plain cURL works as well. Pick whichever your stack already uses; the request shape is the same across SDKs. The dashboard request builder can generate a snippet in the language you select, which is the fastest way to confirm the call works before you wire it into a larger pipeline.
Legality and terms
Public-facing pages can usually be scraped, but review the target site's terms of service before deploying and avoid post-login content. The API does not require a Perplexity account; you sign up for an API key and send the URL. If your use case involves personal data, check the applicable privacy rules for your jurisdiction before storing results.
| Option | Credits per successful request |
|---|---|
| Base request | 1 |
| JavaScript rendering (render_js) | 5 |
| Premium proxy, no JS rendering | 10 |
| Premium proxy with JS rendering | 25 |
| Stealth proxy (stealth_proxy) | 75 |
Figures reflect the published pricing tooltips and FAQ at the time of capture. Plans and credit costs can change; confirm current pricing before committing.
| Plan | Concurrent requests |
|---|---|
| Hobby | 25 |
| Freelance | 50 |
| Startup | 100 |
| Business | 200 |
| Business + | 400 |
| Enterprise 14 | 500 |
| Enterprise 24 | 600 |
| Enterprise 41 | 700 |
| Enterprise 69 | 800 |
| Enterprise 120 | 900 |
Concurrency figures are the published plan values at the time of capture and may change. Confirm current plan limits before sizing a pipeline.
Frequently asked questions
Do I need a Perplexity account to scrape Perplexity News?
No. You sign up for an API key, send the Perplexity News URL, and the service handles the rest. The free tier includes 1,000 credits and does not require a card.
Is scraping Perplexity News legal?
Public-facing pages can usually be scraped, but review the target site's terms of service before deploying and avoid post-login content. Legality depends on your jurisdiction and how you use the data.
What is the rate limit for scraping Perplexity News?
The API applies concurrency limits per plan rather than a per-target rate limit. If the target throttles aggressive patterns, space out requests or use multiple sessions.
How do I get structured JSON from Perplexity News?
Pass extract_rules with CSS selectors mapped to field names, or use ai_extract_rules with a natural-language description of the fields you want. The response comes back as parsed JSON.
Which programming languages can I use?
Any language that can make an HTTP request. SDKs are published for Python, Node.js, Go, PHP, Ruby, and Java, and plain cURL works too.
How much does it cost to scrape Perplexity News?
The free tier includes 1,000 credits with no card required. Paid plans start at $19 per month for higher volume and concurrency. Credit cost per request depends on options: 5 credits for JS rendering, 25 for premium proxy with JS, and 75 for stealth proxy.
Which request option should I start with?
Start with the base request. If the page returns empty or partial content, add render_js. Move to premium proxies or stealth_proxy only when the target blocks the base request, because each step increases the credit cost per successful request.