AI Search - Control AI Search similarity cache freshness
Cloudflare announced new controls for AI Search similarity cache freshness in a changelog post on June 24, 2026. The update allows developers to manage how long similarity search results are cached, providing a configurable time-to-live (TTL) setting. This gives users the ability to balance between serving fresh results and optimizing performance by reducing repeated computations. The feature is part of Cloudflare's AI Search offering, which leverages vector similarity search for AI-powered applications. By exposing cache controls, Cloudflare enables developers to tailor the caching behavior to their specific use cases, such as real-time applications requiring up-to-date results versus those where slightly stale data is acceptable for faster response times. The changelog entry does not specify default TTL values or exact configuration methods, but it indicates that the controls are now available for use. This update is relevant for developers building AI search features on Cloudflare's platform, as it provides more granular management of the trade-off between freshness and latency.
Developers can now tune AI search cache freshness for better performance or real-time results.