Definition
HTTP 429 Too Many Requests is a rate limit response from an API server indicating the client has sent too many requests in a given time window, commonly encountered in search APIs like Gemini, SerpAPI, and Brave when query volume exceeds plan limits.
In Depth
Rate limits protect API infrastructure but can break agent workflows that depend on real-time search. The Gemini API has been returning frequent 429 and 503 errors in May 2026, affecting agents that use Gemini for grounding. Handling strategies: (1) Exponential backoff with jitter -- wait 1s, 2s, 4s, 8s between retries, adding random jitter to avoid thundering herd. (2) Fallback chain -- when the primary search API returns 429, automatically route to a secondary provider. Example chain: Gemini grounding -> Scavio search API -> cached results. (3) Credit budgeting -- track daily spend against plan limits and throttle queries proactively before hitting the wall. (4) Request batching -- combine multiple queries into fewer, larger requests where the API supports it. Scavio's credit-based pricing ($0.005/credit) has no per-minute rate limit on standard plans, making it a reliable fallback when per-minute-capped APIs throttle.
Example Usage
An AI agent using Gemini API for grounding hits 429 errors during peak hours. The developer adds a fallback: on 429, the agent calls Scavio's Google endpoint instead. Retry logic: exponential backoff (1s, 2s, 4s) on the primary, then fallback to secondary after 3 failures. Result: zero user-facing errors during Gemini's outage periods.
Platforms
API Rate Limit (HTTP 429) is relevant across the following platforms, all accessible through Scavio's unified API: