HeadsUpAI

Gemini API Now Lets Developers Mix Built-In and Custom Tools Together

· Updated

The Gemini API now supports combining built-in tools (Google Search, Google Maps) with custom function declarations in a single request — previously developers had to choose one or the other. Three capabilities ship: combined tool support, cross-tool context circulation that preserves each tool's output so follow-up steps can reason over it, and unique tool response IDs (id) for precise mapping in parallel executions. Grounding with Google Maps also expands to the Gemini 3 model family for spatial data, local business info, and commute times.

This removes the orchestration layer agentic workflows needed when hitting multiple sources — public web via Search, private backend via a custom function. Gemini routes between them, cutting latency and simplifying architectures.

Pass built-in tools and custom functions together in one request to build agents that pivot between public data and internal APIs mid-task — use tool response IDs to trace which call produced which output.

View the full update on blog.google
Google AI Developers
Google AI Developers
@googleaidevs
X

Announcing Gemini API tooling updates 🔧 Developers can now combine built-in tools and custom function calling in a single API call for seamless agentic workflows. New tooling capabilities allow devs to: - Combine built-in and custom tools in the same interaction - Cross-tool context circulation for built-in tools - Tool response IDs We’ve also expanded built-in tooling support to include Grounding with Google Maps for the Gemini 3 family of models, enabling models to access up-to-date spatial data, local business info, and more for accurate location-aware responses. Check out the blog to learn more: https://t.co/dB0sg8ye0O

26retweets
View on X

Share this update