Brave Search API
Brave Search API provides web, news, image, video, local, and LLM Context results from Brave's independent index, priced per request with a $5 monthly credit, aimed at AI agents and RAG pipelines rather than Google SERP replication.
What is Brave Search API?
Brave Search API is a developer-facing search service built on Brave's independent web index, not a wrapper around Google or Bing. It is positioned for AI agents, chatbots, RAG pipelines, and applications that need programmatic access to web-scale search results. The product exposes web, news, image, video, local, suggest, spellcheck, and LLM Context endpoints under a single API key and a pay-per-request billing model.
The key architectural distinction is the independent index. Brave crawls and ranks its own results, which means the API does not inherit Google's SERP composition, ranking signals, or availability risk. Brave explicitly markets this as a resilience advantage: Microsoft retired the Bing Search APIs, and Google closed the Custom Search JSON API to new customers, leaving developers looking for alternatives that are not exposed to upstream deprecation. The trade-off is that Brave's index is not Google, and queries where exact Google result fidelity is the product requirement may not be satisfied by an independent index.
How does Brave Search API work?
A request is a standard HTTP call with an `X-Subscription-Token` header carrying the API key. The caller specifies a query and, depending on the endpoint, parameters such as result count, freshness, country, and search language. The API returns JSON containing URLs, titles, snippets, and metadata. The Search endpoint returns conventional web results; specialized endpoints return news articles, image results with dimensions and source URLs, video results, or local points of interest.
The LLM Context endpoint is the product's AI-specific surface. Rather than returning a SERP, it returns extracted passages from ranked pages, formatted for grounding a language model. The caller can set a token budget from roughly 1,000 to 33,000 tokens and choose a relevance threshold (strict, balanced, or lenient). Every source carries a URL and timestamp, and results can be filtered to the last 24 hours. Goggles allow custom re-ranking and domain filtering, and schema-enriched results return structured data rather than plain snippets.
The Answers endpoint adds a synthesis layer: it runs one or more searches and returns a summarized answer with citations, streaming support, and OpenAI SDK compatibility. This is a separate billing unit from Search and carries token-based costs in addition to per-request charges.
Use Cases
- Grounding an LLM agent with fresh web passages through the LLM Context endpoint, without building a separate crawling or extraction pipeline.
- Adding web search to a RAG application where the team wants an independent index rather than a Google or Bing dependency.
- Powering a news monitoring or research tool that needs article results with freshness filtering and source timestamps.
- Serving image or video search results where source URLs and dimensions are required for downstream display or attribution.
- Running local point-of-interest queries for location-aware applications without integrating a separate maps API.
- Building an answer-generation feature with citations where the Answers endpoint's per-request plus token pricing is acceptable relative to assembling the same pipeline from Search calls and an LLM.
Pricing
Pricing model: Paid
| Plan | Price | Billing | Included usage | Notes |
|---|---|---|---|---|
| Search | $5 | per 1,000 requests | Web, news, image, video, local, and LLM Context endpoints. Includes $5 in renewing monthly credit (roughly 1,000 Search requests at $5/1k). Rate limit: 50 queries per second. Goggles, extra snippets, and schema-enriched results included. | — |
| Answers | $4 | per 1,000 requests, plus $5 per million input/output tokens | Summarized answers grounded in single or multi-step searches, with citations, streaming, and OpenAI SDK compatibility. Includes the same $5 monthly credit. Rate limit: 2 queries per second. | — |
| Enterprise | Custom | custom contract | Custom capacity and endpoints, full-funnel Zero Data Retention, custom agreements and NDAs, invoicing, and enterprise-grade support. Contact Brave for pricing. | — |
Common Limitations
- Free credit attribution · The $5 monthly credit is conditional on publicly attributing Brave Search API on the project's website or about page. Removing attribution may forfeit the credit.
- Search result storage · The Search API terms restrict storing search results beyond what is necessary to display them in the customer application. Building a persistent search index or a cached results dataset from API responses is outside the licensed scope. Review the current Terms of Use before designing data-retention pipelines.
- AI training restrictions · Search results and third-party content returned by the API are not licensed for training AI models by default. Separate rights or a custom agreement may be required for model-training use cases.
- Index coverage vs. Google/Bing · Brave operates an independent index (officially described as 30–40 billion pages), not a Google or Bing passthrough. For queries where Google's exact SERP composition or long-tail freshness is the product requirement, the independent index may not replicate Google results. The trade-off is deliberate: no dependency on upstream providers, but not a drop-in Google SERP replacement.
- Standard data retention · Standard self-serve accounts retain search query logs for up to 90 days for billing and troubleshooting. Full-funnel Zero Data Retention is available only on the Enterprise plan. Teams with strict data-minimization requirements should budget for an Enterprise contract rather than assume the self-serve plan is ZDR.
- Answers token cost · The Answers endpoint is billed at $4 per 1,000 requests plus $5 per million input/output tokens. For high-volume answer generation, token costs can accumulate beyond the per-request line item and should be modeled separately.
- Rate limits · The Search plan is capped at 50 queries per second; the Answers plan at 2 queries per second. Higher or custom burst capacity requires an Enterprise agreement.
Key Features
- Web, news, image, video, and local search endpoints
- LLM Context endpoint with token budgeting and relevance thresholds
- Answers endpoint with citations, streaming, and OpenAI SDK compatibility
- Goggles for custom re-ranking and domain filtering
- Schema-enriched structured results
Brave Search API vs. SerpAPI and Tavily
This comparison uses current publicly listed pricing and documented product positioning. Brave is an independent-index search API; SerpAPI is a SERP-scraping provider focused on Google result fidelity; Tavily is an agent-oriented search API with a credit-based model. They solve adjacent but different problems, so the comparison focuses on index source, endpoint design, and cost structure rather than a subjective quality ranking.
| Tool | Positioning | Key differences | Best fit | Pricing note |
|---|---|---|---|---|
| Brave Search API Official pricing and endpoints Privacy notice and retention Search API Terms of Use | A search API built on Brave's independent index, positioned for AI agents and RAG pipelines with LLM-ready endpoints. | Returns results from Brave's own index (no Google or Bing passthrough); offers a dedicated LLM Context endpoint with token budgeting and a separate Answers synthesis endpoint; prices per request at $5/1k for Search and $4/1k plus tokens for Answers; $5 monthly credit is tied to public attribution; standard accounts retain query logs for 90 days, with ZDR only on Enterprise. | Teams that need an independent search index for AI grounding, want to avoid Google/Bing upstream dependency, and can work with a per-request cost structure rather than a subscription. | Search: $5 per 1,000 requests, 50 QPS. Answers: $4 per 1,000 requests plus $5 per million input/output tokens, 2 QPS. Every plan includes $5 monthly credit with attribution. Enterprise adds custom capacity and full-funnel ZDR. |
| SerpAPI SerpAPI pricing | A real-time SERP API that returns parsed search engine results pages, with a strong emphasis on Google result fidelity. | Scrapes Google, Bing, and other search engines rather than maintaining an independent index; returns the search engine result page structure, not extracted LLM-ready passages; pricing starts at $25 per month for 1,000 searches on the entry paid plan, with a free plan of 250 searches per month; cost per 1,000 searches is substantially higher than Brave's Search plan at comparable volumes. | Products that specifically require Google SERP fidelity—rankings, featured snippets, knowledge panels—and have the budget to pay for parsed Google results at volume. | Free: 250 searches/month. Starter: $25/month for 1,000 searches (200/hour). Higher-volume plans scale from there. Enterprise and high-volume tiers are available. |
| Tavily Tavily pricing and free plan | A search API purpose-built for AI agents, with relevance filtering and answer extraction as first-class features. | Offers a free plan of 1,000 API credits per month with no credit card required, then pay-as-you-go at $0.008 per credit; designed around agent workloads rather than SERP reproduction; does not publish an independent web-scale index comparable to Brave's 30–40 billion pages, and its retrieval behavior is optimized for agent consumption rather than raw search coverage. | Agent and RAG prototypes or low-to-moderate-volume production workloads where a generous free tier and per-credit pricing matter more than independent index scale or raw search breadth. | Free: 1,000 credits/month. Pay-as-you-go: $0.008 per credit after the free allowance. A basic search costs 1 credit; advanced search costs 2. No credit card required for the free plan. |
Comparison checked against official product information on 2026-11-08. Features and pricing may change.
Frequently Asked Questions
Is Brave Search API a drop-in replacement for Google Custom Search JSON API?
No. Brave returns results from its own independent index, not Google's. The result composition, ranking, and long-tail coverage will differ from Google. Teams that need the exact Google SERP should use a Google SERP provider (e.g., SerpAPI) rather than treating Brave as a transparent substitute. Brave's value is independence from upstream providers and AI-oriented endpoints, not Google result fidelity.
Is there still a free tier with 2,000 queries per month?
The current self-serve offer is a $5 monthly credit applied to metered billing, which covers roughly 1,000 Search requests at $5/1k. Brave restructured its API plans in February 2026 and replaced the older fixed free plan with this credit model. Legacy subscribers may retain older terms, but new sign-ups should confirm the current plan in the API dashboard rather than assuming a fixed 2,000-query allowance.
Does the $5 monthly credit require attribution?
Yes. Brave states that the free monthly credit requires publicly attributing the Brave Search API in the project's website or about pages. The legacy free plan did not carry this requirement, but the current credit model does. Removing or failing to display attribution may affect credit eligibility.
Can I store Brave search results in my own database?
The Search API Terms of Use restrict storing search results beyond what is necessary to display them in the customer application. Building a persistent search index or cached results corpus from API responses is outside the default licensed scope. If the product requires long-term storage of results, review the current Terms of Use and contact Brave about a custom agreement before designing the pipeline.
Does Brave Search API support Zero Data Retention for all plans?
No. Full-funnel Zero Data Retention is an Enterprise option. Standard self-serve accounts retain search query logs for up to 90 days for billing and troubleshooting. Teams with strict data-minimization requirements need an Enterprise agreement, not the self-serve plan.
Alternative Tools
-
NotebookLM NotebookLM (now Gemini Notebook) is Google's source-grounded research assistant: you upload documents, and it answers only from those sources with inline citations, then turns them into Audio Overviews, Video Overviews, Mind Maps, Slide Decks, and study tools. The free tier is genuinely usable; paid limits are bundled inside Google AI plans. The hard boundaries are no consumer API, answers bounded by uploaded sources, and Audio Overview language quality that is still catching up outside English. -
Tavily Tavily is a web access API for AI agents with a genuinely usable free tier (1,000 credits/month, no card), bundled full-page text in search results, and an official MCP server. But its Research endpoint is dynamically priced at 4–250 credits per request, the headline 1,000 RPM applies only to paid Production keys, entity and academic retrieval lag specialized indexes by large margins, and query data may be used for training unless Zero Data Retention or the data-use setting is changed. -
Elicit Elicit is a systematic review workflow tool with a strong structured extraction interface, but independent evaluations find its search sensitivity averages 39.5% and its quote/reasoning matching degrades to 10%/0% in high-accuracy mode, positioning it as a complement to human reviewers rather than a replacement. -
Exa Exa is a semantic search API built on its own 500B+ page index, with page text included in search results, strong entity-lookup accuracy, an official MCP server, and per-request pricing from $4/1k. It is not a benchmark leader across the board—it is statistically tied with Brave and Firecrawl on independent agentic search benchmarks—and its 10 QPS limit, non-paginating search, crawl-bounded recency, and Enterprise-only compliance features are hard constraints that require workflow design around them. -
Perplexity Perplexity is a cited answer engine that combines real-time web search with advanced language models, but its $20 Pro tier is licensed for personal non-commercial use only, its Pro Search allowance is an unpublished “weekly limit,” and its API is a separate pay-as-you-go billing system. -
Context7 Context7 is an Upstash-built MCP server that injects up-to-date, version-specific library documentation and code snippets into AI coding agents, reducing deprecated-API hallucinations for indexed frameworks and packages.