| Name | Address | Lat, lon | OSM |
|---|
Run the same export from your own code
Every export is a single Overpass query. Below is the one behind the current
export, without the free tool's 5,000 cap. Put your API key in
OVERSPAN_KEY and run it from anything that speaks HTTP. The same
endpoint works with osmnx, overpass-turbo and JOSM; the
quickstart has the settings for each.
[out:json][timeout:60];
area["wikidata"="Q64"]->.a;
nwr["amenity"="restaurant"](area.a);
out center;curl https://api.overspan.dev/api/interpreter \
-H "Authorization: Bearer $OVERSPAN_KEY" \
--data-urlencode 'data=[out:json][timeout:60];area["wikidata"="Q64"]->.a;nwr["amenity"="restaurant"](area.a);out center;' \
-o berlin-restaurants.jsonimport os
import requests
QUERY = """
[out:json][timeout:60];
area["wikidata"="Q64"]->.a;
nwr["amenity"="restaurant"](area.a);
out center;
"""
r = requests.post(
"https://api.overspan.dev/api/interpreter",
headers={"Authorization": "Bearer " + os.environ["OVERSPAN_KEY"]},
data={"data": QUERY},
timeout=90,
)
r.raise_for_status()
print(len(r.json()["elements"]), "features")How it works
-
01
Pick a place
Type a city, region or district and choose it from the list. Places come from Wikidata and are kept only if OpenStreetMap has a boundary for them, so each one means exactly one area. A town with no mapped boundary uses a radius around its centre instead, and the tool says so.
-
02
Pick what to export
One of the common feature types, or any OpenStreetMap tag, such as
cuisine=pizzaorshop=*. -
03
Download the file
The tool runs one Overpass query against a full, minutely updated copy of OpenStreetMap, then builds the CSV and GeoJSON in your browser. Nothing is uploaded and nothing is emailed.
What is in the file
One row per feature. Shops and amenities are often mapped as building outlines rather than points, so those get a single centre point, and every row has a latitude and longitude.
The CSV always starts with osm_type, osm_id,
osm_url, lat, lon and name,
then adds every tag that at least 5% of the results carry, such as
cuisine, opening_hours, website,
phone, wheelchair and the addr:* address
fields. Rare tags would add hundreds of nearly empty columns, so the CSV leaves
them out. The GeoJSON keeps every tag on every feature.
How full those columns are depends on the place. In Berlin, one of the best-mapped cities on earth, nearly every restaurant has a name, about four in five have a street address, and about a third have a phone number. The full breakdown, and what the licence lets you keep, is in our post on building a POI dataset from OpenStreetMap.
The free tool and the API
| Free tool | Overspan API | |
|---|---|---|
| Features per export | Up to 5,000 | No count limit; your plan's memory cap applies |
| Time per query | 25 seconds, about a city or region | 60 to 300 seconds, by plan |
| What you can ask | One tag in one place | Any Overpass QL query |
| How you run it | By hand, on this page | From scripts, notebooks, CI and scheduled jobs |
| Cost | Free, no signup | From $19 a month, cancel any time |
The free tool and the API read the same data. Plans and their exact limits are on the pricing page.
Questions
Q1 Is it really free?
Yes. There is no account, no email address and no key to get. To keep it fair for everyone, each export returns up to 5,000 features, the query gets 25 seconds, and very rapid repeat requests from one address are slowed down.
Q2 How fresh is the data?
It comes from a full copy of the OpenStreetMap planet that follows live edits, usually a minute or two behind. Every export shows the timestamp of the data it read. Identical exports are cached for up to an hour, so a repeat may be up to an hour older.
Q3 Can I use the data commercially?
Yes. OpenStreetMap data is published under the Open Database License (ODbL), which allows commercial use. Credit "© OpenStreetMap contributors" wherever you show the data, and if you publish a database built from it, offer that database under the ODbL too. Using it inside an app, a map or an analysis needs the credit and nothing more.
Q4 Why is a place I know missing?
OpenStreetMap is mapped by volunteers, so coverage depends on how actively an area is edited. Big cities are usually close to complete for shops and amenities, while small towns can be thin. If something is missing you can add it yourself at openstreetmap.org, and it will show up in exports within minutes.
Q5 Why did my export stop at 5,000, or say the area is too big?
Those are the free tool's limits. The count still shows how many features matched. For a whole country, or a query that needs longer than 25 seconds, run the same query with an Overspan API key: per-query time limits go up to 300 seconds and the 5,000 cap does not apply.
Q6 Can I get a shapefile, KML or Excel file?
The CSV opens directly in Excel, Google Sheets and Numbers. For a shapefile, KML or GeoPackage, open the GeoJSON in QGIS, which is free, and use Export, then Save Features As.
Q7 Which tag do I need for something that is not in the list?
Choose Custom tag and type it as key=value, for example
cuisine=sushi, or key=* for any value, for example
shop=*. The OpenStreetMap wiki's
Map features
page lists the agreed tags, and taginfo
shows how often each one is used.
Q8 Is my search stored?
No. The place name you type is sent from our server to Wikidata's public search to find matching places, without your network address. The export runs on our API under the tool's own key. Neither is stored by this website. Results are cached at Cloudflare's edge for up to an hour, keyed only by place and tag. The privacy page has the detail.
Q9 Is this an official OpenStreetMap tool?
No. It is a free tool from Overspan, an independent hosted Overpass API, and is not affiliated with the OpenStreetMap Foundation. The data is OpenStreetMap's, and the credit belongs to its contributors.