Anthropic Launches Prompt Caching Dashboard with Visual Metrics

ME News reports that on April 22 (UTC+8), according to monitoring by Beating, Claude Developer Console has launched the Prompt Caching Dashboard, accessible at platform.claude.com/usage/cache. The dashboard allows filtering by workspace, model, and time period, and primarily displays three metrics: cache read ratio—the proportion of requests that hit existing caches; cache usage composition, which breaks down input tokens into four categories—uncached, 5-minute cache writes, 1-hour cache writes, and cache reads—presented as a stacked bar chart; and write amortization, which measures how many times a single cache write is reused by subsequent reads. In the provided screenshot example, Claude Opus processed 27.4 billion input tokens over seven days, achieving an 85.4% read ratio and a write amortization of 8.65x. At the bottom, there is a time-series chart of cache read ratios, with granularity ranging from 1 hour to 24 hours. Anthropic’s prompt caching mechanism enables API users to mark fixed content such as system prompts and long contexts as cacheable; the initial write incurs an additional fee, while subsequent hits are charged at approximately one-tenth of the standard input rate. Caches are retained by default for 5 minutes, with an option to extend to 1 hour for a fee. Previously, users could only infer caching effectiveness indirectly through token count fields in API responses, without any visualization tools. (Source: BlockBeats)