Google is adding a recent feature for third-party developers, building on Gemini AI models that compete with the likes of OpenAI’s ChatGPT, Anthropic’s Claude, and a growing range of open-source Chinese options that are not likely to be released anytime soon: grounding with Google Maps.
This addition allows developers to mix the inference capabilities of Google Gemini AI models with live geospatial data from Google Maps, enabling applications to provide detailed location-based answers to user queries – equivalent to hours of operation, reviews or the atmosphere of a particular place.
Using data from over 250 million places, developers can now create more intelligent and responsive, location-aware experiences.
This is particularly useful in applications where proximity, real-time availability, or location-specific personalization are essential – for example, local search, delivery services, real estate, and travel planning.
Once the user’s location is known, developers can pass latitude and longitude to the request to improve the quality of the response.
By tightly integrating Maps real-time and historical data with the Gemini API, Google enables apps to generate reasoned, location-specific responses with the factual accuracy and depth of context that is only possible with mapping infrastructure.
The combination of artificial intelligence and geospatial intelligence
The recent feature is available in Google AI Studio, where developers can check out a live demo powered by Gemini Live API. Models that support grounding via Google Maps include:
-
Gemini 2.5 Pro
-
Gemini 2.5 Flash
-
Gemini 2.5 Flash-Lite
-
Gemini Flash 2.0
In one demonstrationa user asked for recommendations for Italian restaurants in Chicago.
The assistant used Maps data to search for the highest-rated options and correct the incorrectly entered restaurant name before locating the correct location with accurate business information.
Developers can also retrieve a context token to embed the Google Maps widget into the app’s UI. This interactive component displays photos, reviews, and other familiar content typically found on Google Maps.
Integration is done through generateContent a method in the Gemini API that developers attach to googleMaps as a tool. They can also enable the Maps widget by setting a parameter in the request. A widget rendered using the returned context token can provide a visual layer alongside the AI-generated text.
Use cases across industries
Mapa’s grounding tool is designed for a big selection of practical applications:
-
Generating an itinerary: Travel apps can create detailed day by day plans with route, time and location information.
-
Personalized local recommendations: Real estate platforms can highlight listings near child-friendly amenities equivalent to schools and parks.
-
Detailed location inquiries: Apps can share certain information, equivalent to whether a café offers outdoor seating, using community feedback and Maps metadata.
We encourage developers to enable this tool only when geographic context matters to optimize each performance and costs.
According to the developer’s documentation, prices start at $25 for 1,000 unreasonable hints – a huge sum for those involved in multi-query trading.
Combining search and maps for higher context
Developers can use Grounding in Google Maps along with Grounding in Google Search in the same request.
While the Maps tool provides factual data equivalent to addresses, opening hours and rankings, the Search tool adds broader context from web content equivalent to news or event listings.
For example, when asked about live music on Beale Street, the connected tools provide venue details from Maps and event times from Search.
According to Google, internal tests show that using each tools together leads to significant improvements in response quality.
Unfortunately, it doesn’t appear like the Google Maps grounding will take current traffic data into account – at least not yet.
Developer customization and flexibility
The experience was built with personalization in mind. Developers can customize system prompts, select from different Gemini models, and configure voice settings to customize interactions.
The demo app in Google AI Studio is also remixable, allowing developers to test ideas, add features, and iterate on designs in a flexible development environment.
The API returns structured metadata—including source links, place IDs, and citation ranges—that developers can use to create inline citations or confirm AI-generated results.
This supports transparency and increases trust in the applications available to the user. Google also requires that Maps-based sources be clearly attributed and associated with the source via their URI.
Implementation notes for AI developers
For technical teams integrating this feature, Google recommends:
-
Passing the user’s location context, if known, for higher results.
-
Displaying Google Maps source links directly below relevant content.
-
Only enable the tool when your query is clearly geographically specific.
-
Monitor latency and disable ground when performance is critical.
Grounding with Google Maps is currently available worldwide, although it is prohibited in several territories (including China, Iran, North Korea and Cuba) and is not permitted for emergency use cases.
Availability and access
Grounding with Google Maps is now generally available via the Gemini API.
With this release, Google continues to expand the capabilities of the Gemini API, enabling developers to build AI-powered applications that understand and respond to the world around them.