Introducing Tool Retrieval in the Gemini API
Introducing Tool Retrieval in the Gemini API
Google launched Tool Retrieval in preview in the Gemini Interactions API, letting agents keep a full tool library on standby and load parameter schemas on demand by adding a tool_search tool and setting defer_loading: true on deferred tools. The model still sees deferred tools' names and descriptions, and when it needs one, the API loads the schema and the model calls the function within the same request; client-side retrieval is also supported for custom registries. Tool Retrieval has no separate charge, works with Gemini 3.6 and later models including all Gemini 4 models, and also supports remote MCP servers.
Useful for developers building large tool-using agents who want to cut input tokens and latency while preserving tool-selection accuracy.
Source: Google Blog AI · blog.google