Brave Search API for AI Agents: How It Works and Pricing
The Brave Search API gives developers and AI apps access to Brave Search, the search engine from the company behind the privacy-focused Brave browser. What sets it apart is that Brave crawls and ranks the web itself: every result comes from Brave’s own index, with no fallback to Google or Bing. For AI, its key product is the LLM Context endpoint, which returns the most relevant passages from the best pages, already extracted and ranked, instead of a list of links. Brave says the API handles hundreds of millions of queries a day and supplies most of the top 10 AI models with web search data.
The quick answer: The Brave Search API is the main independent alternative to Google and Bing for AI search, and it got more important after Microsoft retired the Bing Search APIs in 2025. Its strengths are its own index, low prices, privacy built into the architecture and a large following among open-source agent users. It’s a strong default for grounding chatbots and agents in current information. Everything below comes from Brave’s API page, documentation, blog and about page.
Brave Search API at a glance
Checked against Brave’s Search API page, developer documentation, blog and about page on November 13, 2026.
| Brave Search API | |
|---|---|
| What it is | Web search and AI context APIs from Brave’s own independent index |
| AI endpoints | LLM Context and Answers |
| Other endpoints | Web, news, image, video and place search, autosuggest and spellcheck |
| Pricing | Search $5 per 1,000 requests; Answers $4 per 1,000 plus tokens; free monthly credits |
| Privacy | Zero data retention by architecture; SOC 2 Type II |
| Works with | MCP server, agent skills, OpenClaw and Hermes Agent, OpenAI SDK for Answers |
| Company | Brave Software, maker of the Brave browser, used by more than 120 million people |
Why an independent index matters
For years, most web search APIs either ran on Google or Bing or quietly passed difficult queries to them. That got harder in 2025. Brave’s API page puts it bluntly: “Microsoft retired the Bing Search APIs. Google closed Custom Search API to new customers. Now they’re taking action against scrapers.” In a blog post on the API’s growth, Brave points to Google’s lawsuit against SerpApi, a company that resold scraped Google results, as a warning to anyone building on second-hand search data.
Brave’s pitch is that it doesn’t have this exposure. It calls itself one of only three global-scale search indexes in the West, alongside Google’s and Microsoft’s, and the only one outside Big Tech. The FAQ on its API page is explicit: “Every result comes from Brave’s own independent index, crawled and ranked by Brave. There is no fallback to another search engine for any query.” Brave says this also makes it faster, since long-tail queries don’t take a second hop to another engine.
LLM Context and Answers
Brave’s two AI endpoints serve different needs:
- LLM Context is search for machines. Instead of titles and snippets, it returns the best passages from the best pages, including text, tables, code, forum threads and video captions, ranked in a single call. You can set how much context comes back, from about 1,000 to about 33,000 tokens, choose how strictly it filters noise, filter by date, and use Brave’s Goggles feature to pick which sites ground your model. Every source carries a URL and timestamp. Brave says this endpoint also powers the AI answers in its own search engine, which serve 22 million answers a day.
- Answers returns a finished, cited answer to a question, grounded in one or more searches, with streaming and compatibility with the OpenAI SDK.
The Web Search endpoint, meant for human-style results, plus news, image, video and place search (from an index of more than 200 million locations) round out the API.
Where the data comes from
Brave’s index is built by its own crawler plus anonymous signals from Brave users who opt in to its Web Discovery Project, whose code is open source. Brave argues this produces “less SEO spam, less AI slop, and more of the web that matters,” and its about page says its index uses machine learning and real users’ anonymous contributions. Brave said in its growth post that the index holds about 40 billion pages.
Privacy and zero data retention
Privacy is Brave’s brand, and the API carries it through. Most zero data retention, Brave argues, “is a contractual promise that can only cover queries the vendor answers itself,” whereas Brave’s is architectural because every query ends on infrastructure Brave runs, with no third party to pass it to. Full zero data retention across the whole pipeline is an Enterprise feature, and the API is SOC 2 Type II attested. Storing results or using them to train models requires a custom plan with storage rights.
Pricing
Brave sells prepaid credits rather than subscriptions, and asks for a card only to prevent abuse:
| Plan | Price | Includes |
|---|---|---|
| Search | $5 per 1,000 requests | $5 in free credits each month, full results plus LLM context, Goggles, 50 queries per second |
| Answers | $4 per 1,000 requests plus $5 per million tokens | $5 in free credits each month, citations, streaming, 2 queries per second |
| Enterprise | Custom | Full zero data retention, custom agreements, invoicing and enterprise support |
Brave’s homepage advertises up to $20 a month in free calls. The API also accepts payment from AI agents through stablecoin micropayments instead of an API key.
Benchmarks
Brave takes a skeptical line on benchmarks, writing that “Any benchmark can be gamed,” and points instead to an independent test by the research firm AIMultiple, where it says it ranked first for agent quality and was up to 20 times faster than other search APIs. It also publishes its own runs on FreshQA, a test of questions about current events, reporting 85.3% accuracy at $18 per 1,000 questions including model costs, against 78% to 79% for Parallel, Exa and Tavily at $49 to $89. Those FreshQA figures are Brave’s own; Parallel and Exa publish benchmarks that put themselves on top too.
Adoption
Brave says the API serves hundreds of millions of queries a day, has answered 150 billion in total, and is trusted by “the world’s top AI labs,” including, it says, some it can’t name. Its growth post says it supplies most of the top 10 AI models with real-time web data. It’s especially popular with open-source agents: Brave counts more than a million sign-ups from OpenClaw and Hermes Agent users. Developers connect through an MCP server and official skills for coding agents.
How the Brave Search API compares
The Brave Search API competes with Exa, Tavily, Parallel and Firecrawl. Brave and Exa are the two that most clearly run their own large indexes; Parallel also uses a proprietary index built for agents. Brave’s distinguishing traits are its size and independence from Big Tech, its privacy architecture, simple and low per-request pricing, and its roots in a consumer search engine with real user signals. Exa leans into structured data on people and companies, Parallel into deep research with confidence scores, Tavily into quick setup and safety filters, and Firecrawl into scraping specific sites thoroughly.
If you want an assistant that handles the searching for you, Carly searches the web as part of its work and connects to thousands of apps, so from ChatGPT, Claude or a text message you can ask it to catch you up on a company’s latest news and save the highlights to your notes or CRM.
Who the Brave Search API suits
The Brave Search API is a good fit if:
- You need web search that doesn’t depend on Google or Bing.
- Privacy and zero data retention matter for your users or compliance.
- You want compact, LLM-ready context at a low per-request price.
- You run OpenClaw, Hermes Agent or another open-source agent.
Look elsewhere if:
- You need rich structured data on people and companies.
- You want to store results or train models on them without a custom plan.
- You need deep, multi-step research jobs rather than fast search.
FAQ
What is the Brave Search API?
The Brave Search API gives developers access to Brave Search’s independent web index. It offers web, news, image, video and place search, plus LLM Context and Answers endpoints built for AI apps.
Is the Brave Search API free?
Brave’s Search and Answers plans each include $5 in free credits a month, and Brave advertises up to $20 a month in free calls. Beyond that, Search costs $5 per 1,000 requests.
Does Brave use Google or Bing results?
No. Brave says every result comes from its own independent index, with no fallback to another search engine.
What is Brave’s LLM Context API?
LLM Context is a Brave endpoint that returns the most relevant extracted passages from top pages, ranked and sized for a language model, instead of a list of links.
Is the Brave Search API a replacement for the Bing Search API?
For many developers, yes. Microsoft retired the Bing Search APIs in 2025, and Brave is one of the few remaining providers with its own large-scale web index.
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
