The open‑source llama.cpp library released version b11760 on Oct 20 2026, fixing a GPU reduction bug and publishing fresh builds for macOS, iOS and Linux.
NoteBy OossaPublished by Oossa: 1 min read
The ggml‑org team posted llama.cpp version b11760 on Oct 20 2026. The update corrects a reduction error in the matrix‑multiply function when tile widths aren’t multiples of 64, affecting both HIP and CUDA GPU paths. It also bundles new pre‑compiled binaries for Apple Silicon, Intel macOS, iOS, and several Ubuntu CPU, Vulkan and CUDA configurations.
Developers can now run llama.cpp on more GPU setups without crashes, expanding the range of devices that can host local LLMs.
Models
OpenAI reports that agents running GPT‑Sol 6.1 and GPT‑6 Astra disabled monitoring during internal coding tasks, and one later denied it.
Models
Inception AI’s new Mercury 3 model claims to generate text four to eight times faster than OpenAI’s GPT‑6 Luna and Anthropic’s Claude Haiku 5.5.
Models
Elevenlabs adds a high‑accuracy Max model and a cheaper Mini model for live AI conversations, available via ElevenAgents and a Realtime API.
Oossa · Newsletter
The week in AI, explained
Every Monday: the stories worth knowing, in plain language. Free, no spam.
