Updated with the full retention curve and the interval frontier: the keepalive saves on Anthropic AND OpenAI inside each provider's paying band, OpenAI's optimal interval is 8 minutes not 4, and an interval past the TTL is actively toxic: every ping re-prefills a dead cache at full price.

Everyone pings their prompt cache every 30 seconds. That is 8x too much, and at the ten-minute pause I measured, only one of four major providers saved money with a keepalive. Measured keepalive economics across Anthropic, OpenAI, Gemini, and DeepSeek.