If your Gemini agent connects to dozens of tools, every request starts with a wall of parameter schemas — before your user message even appears. With 200 tools at roughly 400 tokens each, that’s 160,000 tokens of overhead on every turn. Google shipped Tool Retrieval in the Gemini Interactions API on November 19, and the math shifts dramatically: the same agent now opens at roughly 23,650 tokens. That’s an 85% cut from one flag change. How Tool Retrieval Works The mechanism is deferred . Instead of injecting every tool’s full parameter schema into the prompt at the start of each […] The post
Gemini API Tool Retrieval Cuts Agent Token Costs 85%
Google shipped Tool Retrieval in the Gemini Interactions API on November 19, cutting the per-turn token overhead for agents connected to many tools by about 85%. An agent with 200 tools at roughly 400 tokens of schema each drops from about 160,000 tokens of overhead per request to roughly 23,650 tokens, achieved through deferred loading of tool parameter schemas rather than injecting every schema into the prompt at the start of each turn.
Enabling Tool Retrieval in the Gemini Interactions API defers tool schema loading, cutting a 200-tool agent's per-turn overhead from about 160,000 tokens to roughly 23,650 — an 85% reduction from one flag change.
Run your AI side-project on zahid.host
EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.