Google Maps Platform Adds Gemini 3.1 Grounding and AI Models

On April 22, 2026, Google announced a suite of generative AI updates for the Google Maps Platform and Google Earth at the Google Cloud Next conference in Las Vegas. Powered by the Gemini 3.1 engine, the new features shift mapping infrastructure from a standard navigation utility into a spatial data and analytics platform. These updates allow developers to build location-aware generative applications with accurate real-world constraints.

Maps Imagery Grounding

Google introduced Maps Imagery Grounding, currently available in Private Preview for locations in the United States. This tool generates photorealistic AI visuals structurally tied to real-world Street View data. Developers using the Gemini Enterprise Agent Platform can input prompts to render synthetic objects into specific physical locations with accurate geometry and context.

The API also integrates with Veo, Google’s video generation model, allowing developers to animate the static scenes generated by the initial prompt. Advertising agency WPP is testing the API as an early adopter to build immersive, location-based marketing campaigns.

Earth AI Imagery Models and BigQuery Analytics

For geospatial analytics, Google added two new Earth AI Imagery Models to the Google Cloud Model Garden. These experimental models automatically identify specific object classes in overhead imagery, including bridges, roads, and power lines. This eliminates the requirement for engineering teams to train custom object-detection models from scratch for infrastructure planning and disaster assessment.

Google also integrated Aerial and Satellite Insights directly with Google Cloud BigQuery. The automated analysis tools process high-resolution overhead imagery natively. Google claims the tool can compress weeks of manual image review into minutes, specifically targeting city planners tracking residential construction and allocating infrastructure resources.

Infrastructure and Consumer Rollout

The enterprise announcements rely on Google’s newly deployed TPU 8t hardware, optimized for training large-scale models.

The consumer-facing Google Maps application deployed related updates in March 2026, introducing a feature called Ask Maps. This conversational discovery tool operates on Gemini 3.1 Flash Live to handle complex queries with fast, low-latency responses. The consumer application also added Immersive Navigation, a 3D routing experience utilizing Gemini to synthesize Street View and aerial photography into high-detail structural renders of roads, lanes, crosswalks, and traffic lights.

API Access and Pricing Limits

Google Maps currently serves over 2 billion monthly active users. For developers integrating the new generative spatial features, the Google Maps Platform maintains a free tier of up to 10,000 free calls per SKU per month. Production deployments exceeding this volume require the new API subscription tiers introduced alongside the Cloud Next event.

If you build applications requiring geographic context, the Maps Imagery Grounding API changes how you generate synthetic media. You should evaluate the experimental Earth AI models in Model Garden before dedicating resources to training custom object detection pipelines for satellite imagery.