Pricing

Rating

Developer

Actor stats

Categories

Share

Brave Search API

Search the web using Brave's independent search index — not a re-ranked Google or Bing feed. This actor wraps the full Brave Search API v1, giving you structured results for web pages, images, news articles, and videos. Run batch queries, paginate to 200 results per query, filter by freshness and region, and receive clean structured output optimized for LLM pipelines, SEO research, data journalism, and competitive intelligence.

Works out of the box with no API key required (uses zero-config HTML scrape mode for quick tests). For production use, provide a Brave Search API key — the free tier gives you 2,000 queries per month at no cost.

Why Brave Search?

Brave Search maintains its own independent web index, built from scratch without relying on Google or Bing. This means:

• Different results: Brave's index surfaces content that Google's ranking algorithms deprioritize. Great for niche topics, privacy-sensitive queries, and research requiring unbiased results.
• Goggles support: Apply custom re-ranking rules to get niche-optimized results. Goggles let you boost, demote, or restrict results from specific domains or content types — no equivalent exists on Google or Bing.
• Privacy by design: No user tracking, no personalization that skews results. Every query returns the same results regardless of user history.
• Generous free tier: 2,000 queries per month free, then $3 per 1,000 queries. Far cheaper than SerpAPI ($50+/1K) or Bright Data alternatives.
• Independent indexing: Results reflect Brave's independent crawl, not a resold API from big tech.

How to Use

Step 1: Get a free Brave Search API key at https://brave.com/search/api/. The free tier covers 2,000 queries per month with no credit card required.

Step 2: Add your key to the actor input as apiKey. Leave it empty to use zero-config mode (HTML scrape — no key needed, web results only).

Step 3: Provide one or more search queries in the queries field. You can run up to 500 queries in a single actor run.

Step 4: Choose your search type: web (default), images, news, or videos. Configure freshness filters, country, language, and result count as needed.

Step 5: Run the actor and collect structured results from the dataset. Each result includes the URL, title, snippet, source profile, thumbnail, and metadata.

Input Configuration

Common Configurations

Quick web search — default settings, 10 results per query:

{
"apiKey":"YOUR_BRAVE_API_KEY",
"queries":["web scraping tools","best Python libraries 2024"],
"count":10
}

News monitoring — latest articles from the past week:

{
"apiKey":"YOUR_BRAVE_API_KEY",
"queries":["artificial intelligence news","AI regulation"],
"searchType":"news",
"freshness":"pw",
"count":20
}

Image collection — paginate to 100 images:

{
"apiKey":"YOUR_BRAVE_API_KEY",
"queries":["mountain landscape photography"],
"searchType":"images",
"paginate":true,
"maxResults":100
}

LLM context pipeline — markdown output with extra snippets:

{
"apiKey":"YOUR_BRAVE_API_KEY",
"queries":["retrieval augmented generation best practices"],
"outputMarkdown":true,
"extraSnippets":true,
"count":10
}

Localized search — French results for French queries:

{
"apiKey":"YOUR_BRAVE_API_KEY",
"queries":["actualités intelligence artificielle"],
"country":"fr",
"searchLang":"fr",
"uiLang":"fr-FR"
}

Output Format

Each dataset item is one search result. The type field indicates the result kind (web, image, news, video, infobox, discussion, faq).

Web result example:

{
"query":"web scraping tools",
"queryIndex":0,
"resultIndex":0,
"page":1,
"searchType":"web",
"url":"https://example.com/article",
"title":"Top 10 Web Scraping Tools for 2024",
"description":"A comprehensive guide to the best web scraping tools...",
"type":"web",
"age":"3 months ago",
"pageAge":"2024-01-15T00:00:00.000Z",
"profile":{"name":"Example Site","url":"https://example.com"},
"scrapedAt":"2024-06-01T12:00:00.000Z"
}

A run summary is also written to the OUTPUT key of the key-value store with totals for queries, results, API calls, errors, and duration.

Batch Query Use Cases

SEO competitive research: Run 100+ keyword queries in one batch to map the search landscape for a product category. Each query returns up to 20 results (or 200 with pagination), giving you a comprehensive picture of who ranks and what content performs.

Brand monitoring: Track mentions of your brand, product, or competitors across Brave's independent index. Schedule daily runs with freshness: "pd" to catch new coverage.

RAG pipeline enrichment: Pair this actor with a content extractor to build a retrieval-augmented generation pipeline: search → extract full page → chunk → embed → index. Use outputMarkdown: true to get LLM-ready text directly from the search results.

Data journalism: Collect structured search result datasets for a research topic, with timestamps and publication metadata. Use freshness date ranges to capture specific time windows.

Pricing

PPE: $1.00 per 1,000 searches (one event per Brave API call, not per result). Plus your Brave API costs ($0 on free tier, then $3/1K on paid).

Pricing includes all platform compute costs — no hidden fees.

PPE events are charged per successful API call. Failed API calls due to rate limits or server errors are not charged.

Related Actors

• junipr/google-news-scraper — Scrape Google News RSS feeds with topic browsing and keyword search. No API key needed.

FAQ

Do I need a Brave API key to use this actor?

No. Without a key, the actor uses zero-config HTML scrape mode, which works for basic web searches. However, the API mode (with a free key) gives you significantly more data: result metadata, profile information, thumbnails, knowledge panels, discussions, FAQs, and access to image/news/video search types. Get a free key at https://brave.com/search/api/.

What's the difference between Brave Search and Google Search?

Brave Search uses its own independent index — it does not re-rank Google or Bing results. This means Brave surfaces different content, particularly for privacy-sensitive topics or niche queries where big tech search engines apply heavy editorial ranking. Brave also supports Goggles, a unique feature for applying custom re-ranking rules with no equivalent on Google.

How many results can I get per query?

Up to 20 results per API call. With pagination enabled (paginate: true), you can collect up to 200 results per query (Brave's API limit is 10 pages × 20 results). For most use cases, 10–20 results per query is sufficient.

Does this work with Brave Goggles?

Yes. Provide a Goggles URL in the goggles input field. If the Goggles URL is invalid, the actor falls back to standard results and logs a warning. Find community goggles at https://search.brave.com/goggles.

What is the freshness filter and how do I use it?

The freshness filter restricts results to content published within a time window. Use pd (past day), pw (past week), pm (past month), or py (past year). For a custom range, use the format YYYY-MM-DDtoYYYY-MM-DD. Leave empty for no time restriction.

Is there a free tier for the Brave API?

Yes. Brave's free tier includes 2,000 queries per month with no credit card required. Paid plans start at $3 per 1,000 queries. See https://brave.com/search/api/ for pricing.

Can I use this for LLM and RAG pipelines?

Yes. Enable outputMarkdown: true to get clean Markdown-formatted snippets for each result, and extraSnippets: true for additional context. The structured output (URL, title, description, source) maps directly to standard RAG retrieval formats. Combine with a page extractor actor to get full article content.

How is this different from scraping Google Search results?

Using the Brave Search API is fully compliant with Brave's Terms of Service — it's the official programmatic interface for their search engine. Google does not offer a web search API (the Custom Search API is limited to 100 queries/day on the free tier and returns different results than google.com). Brave's API is purpose-built for developers and automation use cases.

What happens if my API key is invalid or quota is exhausted?

The actor detects 401 (invalid key) and 429 quota-exhausted responses immediately, stops processing, and logs a clear error message with the upgrade URL. No PPE events are charged for failed calls.