Reported by Google AI Blog
Tool Retrieval is now in preview in the Gemini Interactions API. Load tool parameter schemas on demand instead of sending your full tool library in every prompt,designed to reduce input tokens and keep tool selection accurate as your agent scales.
We are launching Tool Retrieval in the Gemini API to help your agents manage large tool libraries more efficiently. By loading parameter schemas only when needed, you reduce latency and input token costs for complex tasks. You can try this feature today in the Interactions API by adding a tool_search tool and setting defer_loading to true in your requests.
We are introducing Tool Retrieval to help developers manage large collections of software tools more efficiently. Instead of loading every tool at once, your agent now keeps a library on standby and only pulls in the specific details it needs for a task. This reduces data usage, speeds up performance, and improves accuracy. You can start using this feature today with our latest Gemini models to build more capable and responsive applications.
More from the original report
Today we're launching Tool Retrieval in the Gemini API, available in preview in the Interactions API .
When an agent connects to dozens or even hundreds of tools, sending every full parameter schema on every request adds up fast. It inflates input token counts, adds latency, and lowers tool-selection accuracy as the candidate list grows. Tool Retrieval lets your agent keep its full tool library on standby and load parameter schemas only when the current task needs them.
To turn on Tool Retrieval, add a tool_search tool to your request and set defer_loading: true on the tools you want to hold back. The model still sees the name and description of every deferred tool, so it knows what's available, but the parameter schemas stay out of context. When the model needs a tool, it asks for it by name, the API loads the full schema, and the model calls the function, all inside the same request.
Inside that single API call, the model searches for issue_refund , the API loads its schema, and the model emits the function_call .…
