Every table. Every bed.
260,003 places to eat and sleep across 8 countries — built from open data and public registers, not scraped from listings sites. Free to search, and nothing here has paid to appear.
Choose a country
Every country is built the same way and to the same rules: OpenStreetMap for the venues, public registers for verification, and analysis that has to be computed rather than collected — how dense a strip really is, how a town's mix departs from its national average, which cuisines are genuinely rare where you happen to be standing, and opening hours translated out of machine notation into a sentence you can read.
What you will not find is a rating or a review. We hold none, so we publish none — and we will not infer them, buy them, or lift them from a directory whose terms forbid it. Where venues are ordered, the order is by how much is recorded about them, which is a statement about the data and not about the food.
How a country gets added
Each market starts as a pull from OpenStreetMap — every café, restaurant, takeaway, bar, pub, hotel, motel, hostel, B&B and holiday park that has been mapped in it — which is then deduplicated, assigned to an official subdivision and to its nearest named town, and scored on how much is actually recorded about it. Only then does anything get written.
The details that vary between countries are the ones that matter most. Subdivisions come from ISO 3166-2 codes rather than a bounding-box search, because asking for Canada's provinces by area returns New York, Maine and Minnesota — boundary shapes that merely touch the country come back too. Distance thresholds are per market: a twelve-kilometre radius is a sensible idea of "central" in rural New Zealand and a nonsense in Mumbai or Singapore. And the bar a town must clear before it earns a page is set per country too, because twenty-five venues means something quite different in Tonga than in New South Wales.
Where the numbers stop
Two limits worth knowing before you read anything into a figure here. The first is that this counts venues, not seats, beds or turnover — a district of small cafés will out-count a district of large hotels every time. The second is that coverage follows what people have mapped: a well-surveyed town looks busier than an identical town nobody has walked down. Treat every count as a floor on what exists rather than a census of it, and treat the small countries with particular caution, where a single new listing visibly moves a percentage.
None of that is fixable by us alone, and it does not need to be. Almost every error here is an error in OpenStreetMap, which anyone can correct in a few minutes — and doing so improves every map and app built on it, not just this one. Each venue page links to the record it came from for exactly that reason.
The numbers, by country
| Canada | 89,375 venues · 13 regions · 570 town guides · 5,070 pages |
|---|---|
| India | 78,070 venues · 36 regions · 282 town guides · 758 pages |
| Australia | 64,640 venues · 8 regions · 471 town guides · 1,837 pages |
| New Zealand | 17,649 venues · 17 regions · 190 town guides · 710 pages |
| Singapore | 9,323 venues · 1 region · 85 town guides · 427 pages |
| Fiji | 540 venues · 11 regions · 9 town guides · 6 pages |
| Samoa | 253 venues · 9 regions · 13 town guides · 2 pages |
| Tonga | 153 venues · 4 regions · 6 town guides · 3 pages |
Coverage reflects what has been mapped and published, not what exists on the ground — see how this is built and the open data page.