Updated on
Brave Search: reviews and analysis
A search engine with its own index and AI answers that cite sources, with ads removed for $3 a month.
Affiliate link · no extra cost to you
Our verdict
Brave Search is one of the few AI search engines that does not resell somebody else’s results: its web results have come from an in-house index since April 2023, and the answer layer cites sources as a declared design principle. The pricing split favours the buyer, because searching and asking are free and the $3 a month buys only the removal of ads. It has two serious drawbacks and neither is about price: the index loses to Google on the long tail, to the point where the product offers a fallback that queries Google, and it has carried since 2023 the accusation that its crawler cannot be blocked, which reached court in July 2026.
Best for: Anyone leaving Google on a small budget who wants results that do not come from Google itself.
What people said this week
Tracked from August 3, 2026 to August 9, 2026 · 2 new signals
An industry analysis plays down the effect of that Chrome listing using European regulatory data, and notes the vendor itself attributes part of July's traffic to a football tournament Read it
Brave Search shows up in Chrome's default search engine list, with the caveat that the list is built per country and not everyone sees it Read it
This week’s sentiment is one fifth of the score. It is only computed when there are at least two dated, linkable signals: without them, the score stays on the five sweep axes.
What the internet says
First sweep. Brave Search is one of the few engines in the segment that does not lean on somebody else's index: its own April 2023 announcement states that it removed the last Bing API call, which accounted for around 7 per cent of results, and that every web result has come from its own index since. The AI answer layer is free and cites sources as a principle stated in its documentation, and what costs $3 a month is removing the ads the free tier does carry. The two serious problems are documented by third parties and neither is about price. The first is the long tail, where its own help pages admit in writing that results are not refined enough for every query and offer a fallback that consults Google, available only inside its browser. The second is reputational and dates to 2023, when copyright analysts documented that its crawler does not identify itself in a way a publisher can block, an accusation that underpins a News Corp countersuit filed in July 2026. It is also worth knowing that the index is fed partly by browsing history contributed voluntarily by users of its browser.
What the web repeats in favour
- The in-house index is a verifiable fact rather than a marketing promise, and it is what separates this product from most so-called AI search engines, which are a writing layer over results rented from Google or Bing
- The answer layer cites sources by declared design and is free on top of that, so the part that gives the segment its name is not behind the paywall
- Removing advertising costs $3 a month with no discounts or limited-time labels on the page, a fraction of what any other ad-free search subscription asks
What the web repeats against
- The index falls short on the long tail and the product's own help pages admit it in writing, with a fallback that queries Google as the official fix, available only inside its browser
- The accusation that its crawler cannot be blocked has been open since 2023 and in July 2026 underpins a News Corp countersuit seeking destruction of databases and statutory penalties per infringement
- The free tier carries ads, so the clean-list promise only holds if you pay, and the paid plan has no free trial because the button goes straight to checkout
Sweep sources: Official pricing · Docs · Docs · Docs · Press · Communities · Press
Pros / Cons
Pros
- Its own index for web results, not a layer over Google or Bing
- The AI answers are free and link the sources behind each claim
- Removing ads costs a fraction of what the rest of the segment asks
Cons
- The free tier carries ads, so there is no clean list without paying
- On long-tail queries the index lags behind, and the official way out is falling back to Google inside its own browser
- An open, now litigated accusation that its crawler cannot be blocked
TLDR: Brave Search serves its web results from an index it built itself, and answers questions with links back to the pages each claim came from. The AI answer layer is free. What you pay for, from $3 a month, is removing ads from the results list. It is the cheapest way to leave Google without landing on a product that quietly resells Google. The weak spot is the long tail, where its index still comes up short.
What Brave Search is and how it works
Brave Search is the search engine made by the company behind the Brave browser, and it has one property almost none of its rivals can match: the results come from an index of its own. That sounds like plumbing, so here is why it decides whether the product is worth your attention.
Most things sold today as AI search engines are a writing layer on top of rented results. You ask a question, the product queries somebody else’s search API underneath, and a language model writes the paragraph you read. It works, but the business sits on a supplier who can raise the rate or change the terms whenever it wants. Brave went the other way, cutting its dependence step by step until April 2023, when it announced it had removed the last Bing API call it had left and every web result began coming from its own index.
That decision produces both the appeal and the awkward part. The appeal is that the pages you get back are not Google’s pages in a different typeface. They are a different ordering of the web, built by other people with other priorities. The awkward part is how the index gets built. A large share of it comes from the Web Discovery Project, a system where Brave browser users who switch it on contribute anonymous data about the pages they visit. It is opt-in and it has to be enabled deliberately, but it does mean the index is assembled out of people’s browsing rather than by a conventional crawler walking the whole web.
On top of those results sits the AI layer. When a query calls for it, the engine writes a short answer at the top of the page and links the sources behind each claim, which is precisely the job this segment exists to do. The company frames it as a product principle in its own documentation: the answer arrives with the references that support it so you can judge whether to trust it. A conversational mode came later, mixing search with chat, so you can keep asking about what you just read.
What using it is like day to day
Going by the official documentation and what users report, there is no learning curve. It is a search box that returns a list of links with a written answer on top when the query justifies one. Setting it as your browser default takes two clicks, and after that nobody has to learn anything new.
The difference shows up in the kind of query. On general informational questions, on documentation, on current events, it holds its own comfortably, and the sourced answer saves opening six tabs to reassemble one idea. On deep niche queries, though, the index shows its seams. Technical forums carry reproducible examples of very specific searches that Google resolves in the first result and that surface here on the third page, along with exclusion operators that do not behave the way you would expect.
What is notable is that the product admits this with a feature rather than a footnote. There is a Google fallback option, which has to be switched on and only exists inside Brave’s own browser, that queries the giant privately when the in-house index falls short. Its documentation says in writing that results are not yet refined enough for every query, and points specifically at long-tail ones. The feature is a practical fix and a confession at the same time.
The other daily friction is that the free tier carries ads. They are disclosed, they are plainly formatted, and the company insists they are served without profiling anyone, using only the query, the country and the device type. They are still ads, so anyone who came for a clean results list has to pay.
One more thing belongs here, because it weighs on the decision even though it is not about usage. Since 2023 Brave has carried a sustained criticism about how it obtains and exploits other people’s content. Copyright analysts documented that its crawler does not identify itself in a way that lets a publisher block it selectively, and that its API handed back text snippets far longer than a conventional search excerpt. Three years on, the same accusation is alive and has reached court, with a major publishing group countersuing in July 2026 and repeating the masked-crawler charge. None of that breaks the product for the person using it. All of it is worth knowing if you care where what you read comes from.
Pricing and plans
The split is simple and it favours the buyer. Searching is free. The AI answers are free too, and this is the part people most often get wrong. The only thing that costs money is the subscription that removes the ads, from $3 a month, with no free trial because the button goes straight to checkout. There is an annual option that works out slightly cheaper per month, and the subscriptions page sells each product separately, with no combined bundle carrying a price of its own.
Three dollars is a small number for what it buys, and it is worth placing in context: it is a fraction of what any other ad-free search subscription asks. The right question is not whether it is expensive, because it is not. The question is whether the ad block bothers you enough to pay for its removal when everything else already works without paying.
Watch out for one costly mix-up. The company sells several different subscriptions on the same page, and the one for the browser’s AI assistant is a separate product at a considerably higher price. That fee buys nothing in the search engine. If what you want is a results list without advertising, the search subscription is the one.
Who it is for (and who it is not for)
It suits anyone who wants out of Google without a high monthly fee and without ending up on a product that queries Google underneath. It fits people who search daily for work, people who treat privacy as a requirement rather than a preference, and anyone who wants to check where an answer came from before relying on it.
It is not for people who live on deep niche queries or advanced operator syntax, where the index still loses to the leader and you end up reaching for the fallback. It is not for people whose searches are mostly local businesses and directions, territory where the giant’s maps remain unmatched. And if your problem is not searching the web but talking to an assistant about your own documents, this is neither the product nor the segment.
Alternatives to Brave Search
Inside this segment the natural comparison is Kagi, which answers the same question from the opposite end: no free tier and no advertising at all, you pay from day one, and in exchange you get lenses and per-domain controls over your results. Brave is the option where you start without paying and decide later. Kagi is the option where you pay so you never see a single ad. The full picture is in the best AI search engines, part of AI assistants.
Frequently Asked Questions
Do I have to pay to use the AI answers?
No. The layer that writes answers with cited sources sits in the free tier. What costs $3 a month is no longer seeing ads in the results list, not the artificial intelligence.
Does it really avoid Google’s results?
Web results have come from its own index since 2023, and that is the product’s central argument. The qualifier is that it offers an opt-in fallback that does query Google when its index falls short, so independence is the rule rather than an absolute.
Do I need the Brave browser?
No. The engine works in any browser and can be set as the default. That said, two things do tie it to Brave’s own browser: the index is fed partly by data that browser’s users contribute voluntarily, and the Google fallback only exists inside it. The two pieces lean on each other more than they appear to from outside.
What does it lack against Google?
Depth in the long tail. On very specific queries or ones using exclusion syntax it returns weaker results, with concrete examples documented by technical users. For general and current-events search the gap is much smaller than you would expect.
Is the publisher lawsuit a problem?
For daily use it changes nothing. For your judgement when choosing, it may weigh: the claim that its crawler cannot be blocked has been open since 2023 and reached a courtroom in 2026. If you care whether the tools you use respect what publishers ask for, it belongs in the decision.