llama.cpp Enables Sliding Window Attention for EXAONE 4.5
August 11, 2026
The llama.cpp repository has merged a fix to enable Sliding Window Attention (SWA) for the EXAONE 4.5 model. This update optimizes inference efficiency for models utilizing SWA architectures.
HOW THIS AFFECTS YOU
●
builderYou can now run EXAONE 4.5 more efficiently using SWA support in llama.cpp.