Grounding with Google Maps: Now Available in the Gemini API (blog.google)

Google has added a "Grounding with Google Maps" tool to the Gemini API, letting developers ground LLM responses in Google Maps data (250M+ places) for geospatially aware apps. The model can auto-detect geographic queries or accept latitude/longitude inputs, use Maps sources like addresses, hours, ratings and user reviews, and return a context token that retrieves an interactive Maps widget (photos, reviews, UI-ready details). The tool is available in the Python SDK, supported by Gemini’s latest models, and is generally available now with documentation, pricing, and a remixable demo in Google AI Studio. This matters because grounding provides up-to-date, structured location data that materially improves usefulness and accuracy for travel planning, real estate, retail and logistics. Practical use cases include fully actionable itineraries with distances and travel times, hyper-local personalized recommendations (e.g., family-friendly neighborhoods), and precise place-based answers (e.g., outdoor seating at a cafe). You can also enable Grounding with Google Search alongside Maps—Maps supplies factual, structured data while Search supplies timely, descriptive web context; Google reports substantial quality gains when both are used together. Overall, the feature makes it easier to build interactive, location-aware AI experiences that combine LLM reasoning with authoritative geospatial data.