Local Ai
b9151
Release b9151 of llama.cpp includes enhancements to Unicode regex handling for Qwen3.5's tokenizer, adding a custom non-backtracking handler to prevent stack overflows on long inputs. The release also
Release b9151 of llama.cpp includes enhancements to Unicode regex handling for Qwen3.5's tokenizer, adding a custom non-backtracking handler to prevent stack overflows on long inputs. The release also includes improvements to logging reduction, environment variable fixes, server build fixes, and enhanced prompt processing timings with sampling parameter display.
Source: llama.cpp Releases | 2026-05-14