Local Ai
b9283
Release b9283 of llama.cpp includes updates to the BPE tokenizer with new pre-type dispatch and Carbon-3B DNA tokenizer support, along with refactoring of tokenizer helpers to follow project conventio
Release b9283 of llama.cpp includes updates to the BPE tokenizer with new pre-type dispatch and Carbon-3B DNA tokenizer support, along with refactoring of tokenizer helpers to follow project conventions. The release also fixes a VRAM memory leak in the server context where speculative decoder and draft model resources were not properly freed during sleep/resume cycles.
Source: llama.cpp Releases | 2026-05-22