The Oct 11, 2026 update improves OpenCL performance and enables Gemma‑4 decoding on newer Qualcomm GPUs.
NoteBy OossaPublished by Oossa: 1 min read
The llama.cpp library released version b11559 on Oct 11, 2026. It adds OpenCL improvements for faster factor‑allocation and supports Gemma‑4 decoding on Qualcomm E4B and E2B GPUs. The update also optimizes DK64 and DK128 decode paths for various quantisation settings. The changes were co‑authored by Hongqiang Wang of Qualcomm.
Developers can now run Gemma‑4 models faster on compatible Qualcomm GPUs, expanding edge‑AI capabilities.
Models
Vals AI’s study shows multi‑agent setups can spend up to five times the tokens of a single model for only tiny quality gains.
Models
Odyssey’s Odyssey‑3 model can now generate and run 3‑D worlds from text prompts, with a free online demo and API sign‑ups.
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%.
Oossa · Newsletter
The week in AI, explained
Every Monday: the stories worth knowing, in plain language. Free, no spam.
