Introducing Tool Retrieval in the Gemini API
AI-assisted digest, checked against the original source before publishing. How we work
Google's Gemini API gains a preview feature that lets agents reach hundreds of tools without paying for all of them in every prompt.
Google has introduced Tool Retrieval for the Gemini API, now available in preview. The feature lets developers expose their agents to hundreds of tools while keeping token usage lower than registering every tool up front. By retrieving only the relevant tools for a given request, the API reduces the per-call overhead that grows as agent toolsets expand.
The capability targets builders assembling agents with large, curated tool libraries, where enumerating all available functions in each request would otherwise consume a substantial share of the context window. Google's announcement points to Tool Retrieval as the mechanism for balancing broad tool coverage against prompt efficiency.
Key facts
- • Tool Retrieval launched in preview in the Gemini API.
- • Agents can access hundreds of tools through the feature.
- • Retrieval approach lowers token use versus registering all tools.
Why it matters
Builders can now scale agent toolsets past what fits comfortably in a prompt, shifting the design question from "which tools do I expose" to "how do I retrieve the right ones." Expect tool-selection quality and retrieval latency to become the new tuning surface for Gemini agent stacks.
Spotted an error? Send a correction.
Get the day's AI news in one email
The Daily Epoch · 60-second read · unsubscribe anytime