KV-cache hit rate

termopen source

The share of prompt tokens served from the model's key-value cache; higher means lower latency and cost.

Signals

No signals collected yet.

Relationship map

No mapped relationships yet