Monday, Aug 24 | --:--
Back to home

Google Gemini 3 API Adds Built-in Tools + Function Calling Combination and Maps Grounding

On March 18, 2026, the Gemini API changelog detailed the release of built-in tools and custom function calling combination support plus native Grounding with Google Maps for Gemini 3 models, allowing unified tool use in a single API call.

Tech Insights Reporter 4 min read Mountain View
Cover illustration for Google Gemini 3 API Adds Built-in Tools + Function Calling Combination and Maps Grounding

TLDR

Google released updates to the Gemini API on March 18, 2026, making it possible to use Gemini's built-in tools (such as Google Search and Maps) alongside developers' custom function calling tools within a single API call. Grounding with Google Maps became supported for Gemini 3 models to deliver more accurate location-aware responses.

Built-in Tools and Function Calling Combination

Previously, developers often had to orchestrate built-in tools and custom functions separately. The new combination feature allows both to be specified together in one generateContent request. This reduces complexity for agentic workflows that need both Google ecosystem capabilities (Search, file search, code execution) and arbitrary external or internal functions.

The release notes also reference improved context circulation when using built-in tools, helping the model retain relevant state across multi-step tool interactions.

Maps Grounding for Gemini 3

Grounding with Google Maps is now available for Gemini 3 models. When enabled, responses involving locations, directions, places, or geospatial data can be grounded directly against Google Maps data, providing citations and reducing fabricated details in location-sensitive outputs.

Supported models for the Maps grounding feature are documented in the Gemini API references.

Why this story matters

Reliable tool use and grounding are foundational for production agentic AI. By allowing seamless mixing of first-party and custom tools without extra orchestration layers, and adding real-world map data as a grounding source, Google is lowering the barrier for developers building location-aware, multi-tool agents on Gemini 3. These changes directly address common pain points in current tool-calling implementations.

Sources

Prior Coverage

Earlier Times of AI reporting on this thread.

Scroll to continue reading