Redundancy rate is the share of requests that repeat work the system has already completed. In an AI or web service, a high redundancy rate means the same intent, prompt, query, or lookup is being processed again instead of being reused from cache or another stored result.
This metric matters in cyber security because repeated work can waste compute, inflate cloud costs, and hide abuse. Attackers may exploit redundant requests to drive up API bills, exhaust inference capacity, or create noisy traffic that masks real anomalies. Defenders use exact-match caching, semantic caching, rate limits, and prompt clustering to reduce duplicate processing. Tracking redundancy rate alongside token usage, cache hit rate, and unique outcomes helps teams spot whether the system is doing new work or just paying for the same answer again.