Llama.cpp gains DFlash support

DFlash, a new flash-attention variant, has been merged into the llama.cpp codebase, enabling faster inference for large language models.

One continuously updated timeline instead of dozens of separate articles. New developments are appended as the story evolves.

1. UpdateJun 28, 2026, 01:24 PM 70%

   DFlash flash-attention variant merged into llama.cpp for faster LLM inference

   DFlash, a new flash-attention variant, has been merged into the llama.cpp codebase, enabling faster inference for large language models.

   Read the full story →