The open‑source Llama.cpp library now uses four GDN state columns per CUDA warp, improving GPU performance.
NoteBy OossaPublished by Oossa: 1 min read
The ggml‑org team released llama.cpp version b11494. The update changes the CUDA backend to assign four GDN state columns per warp instead of two. This tweak is the default setting now. Binaries for macOS (Apple Silicon and Intel), iOS, and several Ubuntu builds are available for download.
Developers using llama.cpp on NVIDIA GPUs can expect faster inference without changing their code.
Models
The company’s new three‑agent Claude‑based reviewer found 73.43% of core‑claim errors, far above the previous best of 14.81%.
Models
Mistral's new trillion‑parameter model scores 38 on an independent index, but costs more per task than competing open models.
Models
OpenAI reported two influence campaigns that used AI‑generated content to plant false stories in real media and blocked the involved accounts.
Oossa · Newsletter
The week in AI, explained
Every Monday: the stories worth knowing, in plain language. Free, no spam.
