Google Maps API keys and billing

Is the Google Maps API free?

It is not, but every product has a monthly free cap. The cap varies by tier: Essentials services — dynamic and static maps, geocoding, place details, autocomplete requests and similar — get 10,000 events a month, Pro gets 5,000 and Enterprise 1,000. Usage above that is charged per product.

Much of the content still circulating describes the old model, the universal 200-dollar monthly credit. That credit has been retired, so a budget built on “we're free as long as we stay under 200 dollars” no longer works. In the new model the arithmetic is per product, and exhausting one product's cap does not touch another's.

Do you need a card to get a key?

Yes. A Google Maps Platform key is tied to an active billing account on the Cloud side; even if you plan to stay inside the free cap, no key is created without a valid payment method. This is the detail that leads a developer to start the project on a personal card, with the bill quietly accumulating in the wrong place.

The right setup: the billing account in the company's name on a company card, with the developer granted rights at project level. When the key lives in one person's account, the map stops working one morning after that person leaves.

The three mistakes that blow up the bill

• **An unrestricted key.** A leaked key starts serving someone else's site and the bill is yours. Apply HTTP referrer restrictions on web, package name plus signing certificate SHA-1 on Android, and bundle ID on iOS.
• **No quota ceiling.** A budget alert only informs, it does not stop spending. Set a daily request quota per API; when the limit is reached the calls return errors and spending halts.
• **No session token.** In Places Autocomplete every letter a user types can become a separate request. A session token binds those requests into one search; an app without it bills the same search once per keystroke.

Real ways to bring the cost down

You cannot enlarge the free cap, but you can shrink the request count. The three that pay off most: cache resolved addresses on your side instead of asking again, request data on user interaction rather than on every map load, and render a static image instead of a dynamic map for frequently used fixed points.

Without measurement these turn into guesswork. Cloud reports usage per product; the first job is to see which product the bill comes from. In most teams a single call site produces most of the total, and it is usually not the one anyone expected.

When do alternative map providers make sense?

If the map is used as an image — showing the user a location and nothing more — alternatives save real money. But if the business logic depends on address accuracy, a place database or traffic-aware routing, the comparison belongs on data quality rather than price.

In Türkiye specifically the split shows up here: open data is in good shape in city centres, while address coverage can be thin in newly built areas and rural districts. In a courier, field-team or appointment-address product that gap turns directly into operational cost.