Model techniques map
Techniquesinference & servingKV cache management

general family · filed under inference & serving

KV-cache sharing

Shares KV-cache content to reduce the global cache footprint.

source
1
model
1
lab adopt it
1
strongest
used

How sources treat it

One count per evidence span, weakest treatment to strongest.

used 1

Documented in

Evidence

1 span quoted from the sources, strongest treatment first.

Combined with KV cache sharing [Shazeer, 2019] and the reuse of keys as values in global layers [Kayyam et al., 2026], these optimizations reduce the global KV cache footprint by up to 37.5%.

usedmodel architecturein Gemma 4Google DeepMind

Filed alongside

Other methods under inference & serving :: KV cache management.