semantic caching

token spend control

Cache hits avoid downstream LLM calls, directly reducing token cost and latency.

1 source videos