Cache Stats

SillyTavern does not persist Anthropic (or OpenAI-style) prompt-cache usage on chat messages. Cache Stats captures that usage during generation, stores it on the assistant message, and adds a database icon to the message-actions row.

Tap the icon to see cache hit rate, cache reads, cache writes, uncached input, and output for that reply.

Older messages stay blank until you generate or swipe again because their usage data is no longer available.

Install

In SillyTavern, open Extensions → Install extension and paste:

https://github.com/crossps/st-cache-stats

Enable Cache Stats, then reload SillyTavern. Your next generation or swipe will record statistics. The extension works on desktop and Android installs.

Cache Stats only observes and records usage. It does not change cache duration, provider settings, endpoints, or request behavior.

What it captures

• Anthropic SSE message_start and message_delta usage, including cache_read_input_tokens, cache_creation_input_tokens, and separate 5-minute and 1-hour writes when reported
• OpenAI-style prompt_tokens_details.cached_tokens when supplied by the backend
• Non-streaming usage objects returned as JSON

Quiet and impersonate generations are skipped. Statistics are stored in message.extra.cache_usage and the corresponding swipe extra, so they travel with the chat file.