Brave Place Search API: The Google Maps Alternative That Costs 6–7x Less | Brave
Today we’re releasing the improved Brave Place Search API. It’s a single endpoint for finding places in the physical world (such as businesses, landmarks, and points of interest) drawn from our index containing about 200 million points of interest worldwide, and growing.
This API is the backbone of place search in Brave Search, which handles over 2.2 billion queries a month and powers the Brave Search map. The improved version is available today for public access through the Brave Search API, on the Search plan, at a flat $5 per 1,000 requests.
Two things make the Place Search API worth your attention. First, the quality is comparable to Google Maps, as you will see on the experiment below (which tested a side-by-side comparison of Brave’s Place Search endpoint with 1,000 real queries). Second, Brave’s Place Search API is available at a fraction of the cost of the Google Maps API (which starts at $32 to $35 per 1,000 requests). This makes Brave 6 to 7 times more affordable for comparable place search.
This post details what the API returns, where it’s useful, how it measures up to Google Maps, and why building on Brave’s independent search stack matters.
A single Place Search request returns ranked points of interest with ratings, hours, photos, and distance, on the $5-per-1,000 Search plan.
What it does
One endpoint, accessing an index of about 200 million points of interest worldwide, as well as cities, countries, regions, and streets. This is the backbone of place search in Brave Search, which handles and powers the Brave Search map.
Send a query and a place to look:
curl "https://api.search.brave.com/res/v1/local/place_search?latitude=37.7749&longitude=-122.4194&q=coffee+shops&radius=1000"
-H "X-Subscription-Token: <YOUR_API_KEY>"
No coordinates? Use a name instead, such as location=tokyo japan. Drop the query entirely and you get Explore mode: a snapshot of what’s around a point, ideal for map views. It even works without any location hints at all, doing a global search instead.
Every result comes back ready to render:
- Name, URL, coordinates, and full postal address
- Ratings, review counts, price range, categories, and cuisine
- Opening hours (including today), phone, email, and timezone
- Photos and distance from your search center
All these attributes are available on a flat rate of $5 per thousand requests. Compare that with Google, which charges you more for additional attributes.
Need deeper data? Every result carries an id. Pass it to /local/pois for more detail or /local/descriptions for AI-written summaries. The best part: those IDs also come from Web Search results, so one integration covers your local and Web surfaces at once.
Useful in multiple ways for apps and agents
You know your product better than we do. So we won’t tell you what to build. But the data is rich enough that one call covers jobs you’d otherwise wire up several APIs to handle:
- “Near me” discovery: restaurants, gyms, EV chargers, ATMs. Anchor to coordinates, set a radius, render a ranked list.
- Travel guides: explore mode plus the
citiespanel shows what’s worth seeing, then drills into hotels and attractions. - Business directories: try “find a location near you” functionality without licensing a maps stack.
- Map dashboards:
zoom_levelhints and coordinates drop straight into your tiles. - Geofenced nudges: try searches like “three highly-rated lunch spots within 500m.”
- AI agents: structured, current place data a model can reason over.
The point is the breadth. Where you take it is up to you.
One Place Search response, four surfaces: a “near me” list, a travel guide, a map view, and structured context for an AI assistant.
Note the radius parameter will bias toward an area rather than hard-cutting at the edge. Under ~20 km it gives tight “near me” results. For a famous landmark, go wide or skip it. Match it to the job and results stay clean.
How it compares to Google Maps
Google Maps is the more mature product, and we won’t claim otherwise.
But the question that matters is whether Brave is good enough to build on, and what you gain by choosing it. To answer it, we ran 1,000 real, global, and multilingual place queries through both Brave and Google Maps. An LLM judge scored every pairing on recall, precision, ranking, and overall quality.
The result: a close second in overall quality: 6.4 vs 7.3 out of 10.
The two engines aren’t tied in character, though. They win in opposite places. That’s the useful part.
Where Brave wins:
- Coverage: Brave returns more of the places that actually exist (recall 7.2 vs 6.8). Google often returns fewer.
- Ambiguous names: Search `